du intof matica ^
YU ISSN 0350-5596
informatica
Published by Informatika, Slovene Society for Informatics, Parmova 4 1, 6.1000 Ljubljana, Yugoslavia
Editorial Board
Suad Alagić, Sarajevo; Damjan Bojadžiev, Ljubljana; Jozo Dujmović, Beograd-, Janez Grad, Ljubljana, Bogomir Horvat, Maribor; Ljubo Pipan, Kranj; Tomo Pisanski, Ljubljana; Oliver Popov, Skopje; Sašo PreSern, Ljubljana; Viljem Rupnik, Ljubljana; Branko Souček, Zagreb
Editor-in-Chief : Prof. Dr. Anton P. Zeleznikar
JOURNAL OF COMPUTING AND INFORMATICS
YU ISSN 0350-5596
VOLUME 12, 1988^ No. 4
Executive Editor : Dr. Rudolf Murn
CONTENTS
Publishing Council:
T. Banovec, Zavod SR Slovenije za statistiko, Vožarski pot 12, 61000 Ljubljana;
A.	Jerman-Blažič, Iskra Telematika, Kardeljeva
ploščad 24, 61000 Ljubljana;
B.	Klemenčič, Iskra Telematika, 64000 Kranj;
S. Saksida, Institut za sociologijo Univerze Edvarda Kardelja, 61000 Ljubljana;
J, Virant, Fakulteta za elektrotehniko, TržaSka 25, 61000 Ljubljana.
Headquarters :
Informatica, Journal for Computing and Informatica, Iskra Delta Computers, Stegne 150,
61000 Ljubljana, Yugoslavia Phone: (+38 61) 57 45 54. Telex: 31366 yu delta Fax: (+38 61) 32 88 87 and (+38 61.) 55 32 61.
Annual Subscription Rate: US$ 30 for companies, and IIS$ 15 for individuals.
Opinions expressed in the contributions are not necessarily shared by the Editorial Board
Printed by: Tiskarna Kresija, Ljubljana
A.P.Zeleznikar
B.	Dugonik
I. Bruha
M. Stojanović I. Stojmenović
J. Barle J. Grad
R.	AndrejCiö
B.	Peček
M.	Ivanović
Z.	Budimac
S.	Bobek
B. Domanjko A,P.Zeleznikar
3 Informational Logic II
21 Temporal Logic - A Tool for System Modeling
25 Compact Implementation of the Programming Language Prolog in the Environment of Mc Poplog
36 On Deletion of Reflex Elements from Circular Doubly-Linked Lists
39 Computer Program for Determining an Optimum Solution in Long-Term Forest Exploitation Process
44 Source Code Analysis
49 Translation of Grammar Rule EBNF Descriptions to BNF Ones
55 Quo Vadis Informatization in Business Decision Making
59 UUCP Network
63 Information and Information System
69 Forum Informationis
informatica
Časopis izdaja Slovensko društvo Informatika, 61000 Ljubljana, Parmova 41, Jugoslavija
ČASOPIS ZA TEHNOLOGIJO RAČUNALNIŠTVA
IN PROBLEME INFORMATIKE
ČASOPIS ZA RAČUNARSKU TEHNOLOGIJU I
PROBLEME INFORMATIKE
SPISANIE ZA TEHNOLOGIJA NA SMETANJETO
i PROBLEMI OD OBLASTA NA INFORMATIKATA
Uredniški odbor:
Suad Alagič, Sarajevo; Damjan Bojadžiev, Ljubljana; Jožo Dujmovič, Beograd; Janez Grad, Ljubljana, Bogomir Horvat, Maribor; Ljubo Pipan, Kranj; Tomo Pisanski, Ljubljana; Oliver Popov, Skopje; Sašo PreSern, Ljubljana; Viljem Rupnik, Ljubljana; Branko Souček, Zagreb
YU ISSN 0350-5596
LETNIK 12, 1988-ŠT. 4
Glavni in odgovorni urednik: prof. dr. Anton P. Zeleznikar
Tehnični urednik: dr. Rudolf Murn
Založniški svet:
T. Banovec, Zavod SR Slovenije za statistiko, Vožarski pot 12, 61000 Ljubljana;
4.	Jerman-Blažič, Iskra Telematika, Kardeljeva
ploSCad 24, 61000 Ljubljana;
B. Klemenčič, Iskra Telematika, 64000 Kranj;
5.	Saksida, Institut za sociologijo Univerze
Edvarda Kardelja, 61000 Ljubljana;
J. Virant, Fakulteta za elektrotehniko, Tržaška 25, 61000 Ljubljana.
Uredništvo in uprava:
Časopis Informatika, Iskra Delta, Stegne 150, 61000 Ljubljana, telefon (061) 574 554; teleks 31366 YU Delta; fax (061) 328 887 in (061) 553 261.
Letna naroCnina za delovne organizacije znaša 24000 din, za zasebne naroönike 6000 din, za študente 2000 din; posamezna številka 8000 din.
Številka žiro računa: 50101-678-51841
Pri financiranju časopisa sodeluje Raziskovalna skupnost Slovenije
Na podlagi mnenja Republiškega komiteja za informiranje št. 23-85, z dne 29. 1. 1986, je časopis oproščen temeljnega davka od prometa proizvodov.
Tisk: Tiskarna Kresija, Ljubljana
VSEBINA
A.P.Zeleznikar
B.	Dugonik
I. Bruha
M. Stojanovič I. Stojmenovič
J. Barle J. Grad
R.	Andrejčič
B.	Peček
M.	ivanović
Z.	Budimac
S.	Bobek
B. Domanjko A.P.Zeleznikar
3 Informational Logic II
21 Temporal Logic - A Tool for System Modeling
25 Compact Implementation of the Programming Language Prolog in the Environment of Mc Poplog
36 On Deletion of Reflex Elements from Circular Doubly-Linked Lists
39 Computer Program for Determining an Optimum Solution in, Long-Term Forest Exploitation Process
44 Source Code Analysis
49 Prevodjenje pravila gramatike iz EBNF zapisa u BMF.zapis
55 Quo vadiš informatizacija poslovnega odločanja
59 UUCP mreža
63 Informacija in informacijski sistem
69 Forum informationis
INFORMATIONAL LOGIC II	INFORMATICA 4/1988
Anton. P. Železnikar
UDK 519.72	Iskra Delta, Ljubljana
This part of the essay deals with the formal informational logic, however only with a part of this logic, i.e. with the definitions of informational variables and operators which constitute the so-called list of primitive symbols. In the continuation of this essay formation rules, inTorraational axioms, and informational transformation rules will be discussed in detail. The topic of this part of the essay is the following: Introduction into Formal Informational Logic and Basic Informational Variables and Operators. In the introductory section the syntax for labelling definitions, theorems, proofs, examples, and informational systems is exposed. Then, in the second section, various definitions of informational variables are introduced and three types of equivalence operators, used in the essay, are explained thoroughly. The next subsection is dedicated to general informational metaoperators , which are meant to be operational variables. These variables can be particularized and universalized into concrete operators (e.g., through a non-uniform substitution in a formula). Thereupon, operators concerning the informational cycle are determined. Within these, counter-informati ona1 operators and operators of embedding are emphasized. Operators of parallel Informing are discussed in a most exhaustive way. There are operators of general parallel Informing, parallel informational possibility and necessity, transfer of information among parallel processes, informational appearance and vanishing, informational choice, memorizing, forgetting, and renewing of information, informational interrupting and breakdown, and enriching of information. For better understanding, several examples of formulae including various operators are given. As it has already been mentioned, in the continuation of the essay, in Part Two, formation rules, informational axioms, and informational transformation rules will be included.
Informacijska logika II. Drugi del: formalna informacijska logika. Ta
del spisa obravnava formalno informacijsko logiko, toda le del te logike, in sicer definicije informacijskih spremenljivk in operatorjev, ki oblikujejo t.i. seznam primitivnih simbolov. V nadaljevanju spisa bodo obravnavana podrobno še formacijska pravila, informacijski aksiomi in informacijska transformacijska pravila. Naslovna tema tega dela spisa je uvod v formalno informacijsko logiko in osnovne informacijske spremenljivke in operatorji. V uvodnem poglavju je opisana sintaksa označevanja definicij, teoremov, dokazov, primerov in informacijskih sistemov. V naslednjem poglavju se uvaja več definicij informacijskih spremenljivk in trije tipi ekvivalenčnih operatorjev s pojasnili njihove uporabe v tem spisu. Naslednje podpoglavje je namenjeno splošnim informacijskim (meta)operator jem, ki se razumevajo kot operacijske spremenljivke, ki jih je mogoče konkretizirati (jim prirejati vrednosti) s t.i. partikularizacijo in univerzalizacijo (npr. z neuniformno substitucijo v informacijskih formulah) . Potem se opredeljujejo opera"torji informacijskega cikla. V okviru teh so posebej obravnavani proti informacijski in vmestitveni operatorji. Operatorji paralelnega informiranja so najštevilnejši. To so operatorji splošnega paralelnega informiranja, paralelne informacijske možnosti in nujnosti, prenosa informacije med paralelnimi procesi, informacijskega nastanka in izginotja, informacijske izbire, pomnenja, pozabljanja in obnavljanja informacije, informacijskega prekinjanja in preloma in o'bogačevanja informacije. Z namenom boljšega razumevanja spisa je konstruiranih več primerov formul z različnimi operatorji. Kot je bilo že omenjeno, bodo v nadaljevanju tega spisa, v njegovem naslednjem delu opredeljena še formacijska (oblikovalna) pravila formul, informacijski aksiomi in informacijska transformacijska pravila formul.
Part Two: Formal Informational Logic
the necessary systematics and uniformness.
II.0. INTRODUCTION
TI. 1. SOME BASIC INFORMATIONAL VARIABLES AND OPERATORS
Logical thinking is an improvement of na tura 1 thinking because it trims the exuberance of natural thinking. Logical thinking is a deliberate attempt to restrain the excesses of natural thinking. This restraint is effected by selectively blocking natural flow pathways.
Logic is the management of NO.
Edward de Bono [8] 224
In the first part of this essay a general and non-systenitit i c (intuitive) view of i nfonniition-al logic and its possibilities were shown, dealing principally with the philosophy and conceptualism of informational formalization through introduction of a new seniuntios for basic logical operators. In this, second part of the essay, an informationally axiomatic approach will be given, constituting the so-called symbolic system of informational logic.
For the purpose of ciarity of the new formalism, we will introduce several symbols for operators with already known, but also new semantics. Furthermore, we will use particular forms of marking and presenting definitions, theorems, proofs, and examples. Among these formalistic parts of the essay we shall put explanatory inserts for making formal achievements more picturesque and '' comprehensible. In a formal as well as in an intuitional manner, we shall develop and reveal a novel understanding of the arising informational formalism which, to our opinion, can constitute the possibility of today's and probably also of tomorrow's informational investigations and understanding.
To have a good perspicuity, we shall use the following syntax for labelling definitions (OF), theorems ( TH) , their proofs (PR), examples (EX), and informational systems (SY):
DFl
[Syntax of labelling] :
<type>(number)
<label> ::=
[<characteristic_formula>] <type> : := DF|TH1 PR|EX 1SY <number) ::= 1|2|3|...
Definitions will begin by their label at the left of their first line and will end with the mark	Theorems will begin by the label in
the middle of their first line and will end with the mark ' ■ 'Proofs, examples, and informational systems will begin by the label at the right of their first line and will end with the mark	. Generally, texts of
definitions, theorems, proofs, examples, and informational systems will follow the general syntax
<label>
<text_of_the_type> This order of labelling will bring to the text
The battle of reason is the struggle to break up the rigidity to which the understanding has reduced everything.
G. W. F. Hegel [9] 53
In this section (II. 1.) we shall try to create a sufficiently broad basis of semantics of informational variables and operators, so that in the subsequent sections of this essay it would be possible to construct the axiomatic basis and the mechanisms of constructing informational formulae by means of proposed formation and transformation rules. The semantic basis of this section is relevant for it can substantially influence the possibilities of arising of informational logic. In this illumination we have to develop the semantics of informational variables and particularly of informational operators, considering the limits of the intuitive concept of information within the attainability of the human informational mind.
As in some recently discussed cases, we should like to distinguish clearly the variables concerning information, and those concerning Informing, Although information is in no way a passive informational component, in some expressions it will be put comprehensively in a passive state, and it will function as an operand. In this state information will be denoted by a small Greek letter. However, it is not to say that information will not influence other information or that it will not function as an implicit operator. Informing as an informational operator over information will be denoted by a capital Latin letter. However, it does not mean that Informing is not under the influence of other Informing or information, that is that it does not function as an (implicit) informational operand. Except of these two types (categories) of information (called information and Informing) we shall have some distinguished (logical) operators, considering the specifically investigational or informationally principal properties of information. These distinguished operators will constitute the logical or philosophical background of the presented informationally logical discourse.
II.1.0. Informational Variables
DFl
[Variables] :
Variables, presenting information in, cases of operands and operators, will be denoted by distinguished small Greek or capital Latin letters and operational symbols. In cases of general information representing informational operands as variables, letters a, ß, ... , oj (non-indexed and indexed) will be used. Exceptionally also identifiers written in normal or Italic will be useful for representation of informational operands and
operators. Informational cases corresponding to particular forms and processes (e.g. counter-Informing, counter-information, embedding, information of embedding, metaphysics, etc.) will be marked by distinguished letters in indexed and non-indexed form.	■
DF2
[Variables] :
Variables, representing information of a living being (cell, organism, etc.), will be, for instance, (ji for metaphysics, cr for sensory information, A for motpr information, t] for emotional information, etc. Exceptionally also identifiers written in Italics, will be used. Further, 7t will be \ised for marking the possibility, V for the necessity and for the net of informational traces, X for marking informational lumps, t for informational traces, etc.	■
ÜF3
[Variables]
Variables, presenting processes of Informing or Inforraings as operand variables, will be denoted by distinguished capital Latin letters. In the cases of general Informing, I (non-indexed and indexed) will be used. Exceptionally also identifiers written in Italic wilJ be used. To Informing I corresponding counter-Informing and informational embedding will be denoted by C and E (non-indexed and indexed), respectively.	■
DF4
[Variables] :
Variables and operators, presenting processes of Informing of a livinj^ being, wi .1 .1 be (ienoted by M, S, L, Y, etc., for metaphysical Informing, sensory Informing, motor Informing, emotional Informing, etc., respectively.	■
DF5
[Variables] :
The general variable representing any operator of informational logic will be the symbol (= and other distinguished" symbols declared as informational operational variables. These variables are representatives of generative sets of all (also the imaginative ones) possible informational operators. The values of these operational variables will be subscripted or superscripted operators of the type |= or similar operator symbols or a semantically determined common symbols (e.g.. A, V, , =, C, X, n, U, o, +, =, etc.). It is to stress that informational logic introduces the institution of operators being informational variables in a similar sense as it introduces variables representing possible operands. In some cases, operational variables (markers for operators denoting variables) will be called metaoperators. The direction of the operation of an informational operator can be subscripted by (from left to right) and by (from right to left).	■
II. 1.1. Types of Equivalence Operators
■Equivalence operators concern the opposite of the difference which is the most natural phenomenon of information. Informational equivalence (relations) must be considered to be purely theoretical, that is symbolical. Materialized, processible, or phenomenological cases of the so-called strict equivalence do not exist in nature; they can exist solely as informational phenomenon of abstraction, idealization, or, in general, as a being's metaphysical imagination.
[Operators] :
Any concrete equivalence operator represents an operational value of the operational variable . Of course, we can also have an equivalence operational variable marked by	, for
instance. In our discourse only three different types of equivalence operators will be used, namely,
for definitional equivalence,
'='	for symbolic equivalence, and
'=' for logical equivalence.	■
The definitional equivalence always introduces semantics of a new symbolism. The symbolic equivalence introduces (in fact, by definition) simply a new symbol, for instance, a shorter or a more adequate one. The logical equivalence concerns the classical logic (e.g. the truth and falsity) in informational well-formed formulae (iwffs).
II. 1.2. General Informational Metaoperators
... we seek approved or verified definitions, the content of which is not assumed merely as given, but is seen and known to warrant itself, because warranted by the free self-evolution of thought.
G. W. F. Hegel [9] 146
DF2
[Operators]
Two general metaoperators (informational, however operational variables) of Informing will be introduced. Let us have the following definition:
N V =1
"Df
'informs'	V
' inform'	V
'isinformed' V 'are informed'
Although variables representing information and Informing will be passive and active or will have the roles of operands and operators, some especially distinguished information-logical or information-theoretical operators will be introduced into the "logical" discourse which follows. Some of these operators will be semantically broadened to satisfy also the requirements of the most general informational circumstances (informational phenomenology).
In this case, \= and =j are understood to be unary, binary, or multiple prefix, postfix, or infix operators (as it will be presented later in several oases), which generally concern operand, operator or mixed information (for instance informational well-formed formulae within a logical system of information). In a general.case (the so-called universal case) the operator f= will incorporate also the meaning of the operator H and vice versa. Thus, the usage of the operator h will denote not only the meaning of both of them but also an arbitrary
particular operational meaning. It is to say that operators (= and H will concern a general case of Informing as well as the so-called cyclical Informing or an integration of several informational cycles. In a similar way, they will represent also the so-called parallel informational cases of Informing. According to their application, needs, and the goals of an informational scenario, these metaoperators will be particularized {informat i ona11y specialized) by using subscripts and universalized ( informationally generalized or variegated) by using superscripts. It is to say that operators |= and =| are symmetrical with regard to the direction of operation, for instance, from the left to the right and vice versa, respectively. It would be quite sufficient to introduce only one of them and determine the second one as a particularization of the first one.	■
EXl
tOperators] : The introduction of two operators t= and instead of only one of them enables the expression of a parallel process in a single linear form. For instance, instead of two parallel processes a t= ß and ß H T> it is possible to write simply a ß ^ t . In this case ß is informed by a and simultaneously by T-	■
DF3
[Operators] :
As mentioned in the previous definition, we can define the following two operators with their directionally different meanings:
¥ ^ A
Df

Df
Df
3(|=,	A H)
3(=i, t=).(=l A 1=)
and
where these expressions are read as "there exist operators ... and ... such that (.) ...". A is the relation of dominance. In the first definition, operator ^ is the so-called feedback operator (or to |= subordinated operator) and so is in the second definition. In both cases, the relation of dominance A is not commutative. The feed-back operator is in fact functionally weaker than the main operator, so, it is put into the superscript position. We introduce the so-called operator (or relation) of the dominance of Informing A, where for the first and for the second case
1= A =(' and =| A t=
were valid, respectively.	I
DF4
[Operators] :
The operators	and	in the previous
definition can be understood as universalizations of the operators t= and =j- In these cases it would be meaningful to introduce the superscript mark of universalization » in the following way:
Df
.DF5
and
Df
[Operators]
DF2
In opposition to [Operators]	two general
metaoperators (informational variables) of non-Informing can be introduced. Let us have the following definition:
'does_not_inform' v 'do_not_inform' v 'is_not_informed' V 'are not informed'
It is to understand that ^ and 4 are unary, binary, or multiple prefix, postfix, or infix operators concerning a general, a cyclical, as well as a parallel informational case. It is to stress that a particularization of these operators does not mean a non-Informing for cases, which do not concern particularization. Thus, non-Informing really remains only particular.	■
for instance has the meaning of a general
non-Informing. The more universally the
operator [= is determined, the broader is the
domain of non-Informing of the operator fet. A
particularization of localizes the non-
Informing to a particular domain. In the case
of for instance, the domain of non-Informing B
is narrowed to the domain of belief (the subscript B). In classical logic, can have
the meaning of the non-true Informing, i.e. of the false Informing. In this respect becomes logically equivalent to	etc.
DF6
[Operators] :
DF3
In a similar way as in [Operators]	four
additional (universalized) operators of Informing can be defined, for instance:
	^Df	3(t=,		.(N A
	^Df		=i)	A H)
	"^Df		f )	.(=i A
	^Df		1=)	.(H A N)
The short form notation of these operators
/♦ ♦ /♦ ♦
could be, for instance, [= , fe^ , H » and ^ , respectively.	■
As it can be understood from these four cases, a universalization means also absolutizing in the sense of Informing. In the last four cases universalization has in fact been unionization (for instance, in set-theoretical sense) of two informational operators, which becomes evident from the following definition:
[Operators] :
The universalization of two operators <opj^> and <op^> can be achieved by defining the new operator in the following way:

Df
<oPj> <op2>
where the defined universalized operator can be marked by
<op^>
<s>^
<0P2>
<s>^
where <s> is the syntax variable of symbols and the syntax variables <op^> and (oPg) are
markers for generative classes (types) of informational operators. The disjunctive connective V^^ emphasizes the parallelness of
<op^) and/<op2>. Additionally, <op^> A <op2>
can be defined.
DF8
[Operatora]
Now, it becomes evident that also universalized operators of explicit non-Informing can be determined. According to the previous definitions of operators, the following two cases can be distinguished as consequently the non-Informing ones:
Df
Df
A) A H) and u{H, t/). (H A bt)
/* /*
These operators can be denoted as Y find > respectively.	-	E
II. 1.3. Operators Concerning the Informational
Cycle
Informational cycle needs to be additionally explained form the point of view of semantics. This cycle is under-stood to be an informational subprocess with some characteristic basic processes. In general two basic processes or characteristic cyclical sub-Informings can be distinguished; the so-called counter-Informing and embedding of the arisen -information called counter-information. These two processes constitute the so-called Informing of the informational cycle. In this respect, similarly to the case of general Informing, two additional (generally cyclical) metaoperators can be introduced.
I)K9
[Operators] :
Two general operators (operational variables), explicitly concerning the informational cycle, can be introduced in the following way:
h V H
Df
'informs_in_a_single_
inforraational_cycle' 'inform_in_a_single_
informational_cycle' 'is_informed_in_a_single_
informational_cycle' 'are_inf ormed_in_a_single_ inf ormational_cycle'
Of course, the cyclical operators f- and H can be understood to be subclasses of the general operators t= and =t, respectively. Also, the operator h can be universalized to incorporate the meaning of the operator H, and vice versa. It is to stress that the operators |- and -| concern explicitly the so-called cyclic Informing. If these two operators are particularized, the par t iculari zat i on can concern only the so-called cyclical nature of Informing. Thus, each particularized or universalized operator of the type H or H concerns cyclical or. intercyclical (cyclical) Informing.	■
Let us show and discuss some recursive defi.nitions of cyclical operators and the meaning proceeding from these definitions.
DF3
[Operators]
Similarly to the [Operators] , it is possible to determine in a recursive way two bidirectional (universalized) operators, namely:
H
Df "^Df
3(1-, H).(|- A H) 3(H, h).(^ A h)
where these expressions are read as "there exist two cyclical operators ... and . . . such that ( . )■ ...". In the first definition, operator H is the so-oallod cyclical feed-back operator (or to h subordinated cyclical opiu-ator) litici so i h |- in the second definition. As |- and H are already cyclical, the operators
|- ' and	in fact concern intercyclical
processes (as cycles between two cyclical processes). However, cycles can include sub-cycles in a serial and parallel form, thus, cyclical (and non-cyc1 ica1 ) operators can appear without hesitation within informational cycles. Similarly to the definition DF4
[Operators] , these two general intercyclical operators can be denoted in their universalized form in the following way:
Df
h
and
Df

DFll
[Operators]	:
In opposition to the [Operators]	, according
DF5
to the [Operators] , two cyclically general metaoperators (informational variables) of cyclical non-Informing can be introduced. So, let us have the following definition:
M V
Df
'does notinform_cyclically'	V
'donotinformcyclically'	v
' i s riot i nf ormed_cycl ical ly '	V 'are ..not inf ormedcycl ically '
If Informing is not cyclical, it can still be the general, parallel, or non-cyclically particular one. In this respect, the cyclical non-Informing is an informational particularity too, concerning the concept of informational cycle.	■
The semantics of the informational cycle has been constituted through the introduction of two characteristic cyclical informational processes: counter-Informing ^and embedding of arisen counter-information. Within the cycle, the counter-Informing is' the so-called progressive (informationally generative) process, in which counter-information arises. The embedding of arisen information is the so-called regi-essive (informationally connective) process, which is^necessary to keep the arisen counter-inforraat i on preserved for future Informing. The informational cycling becomes possible through the circulation of progressive (counter-informational) and to these processes following regressive (informationally embedding) processes. However, in this circular or cyclical process, counter-Informing and embedding are cyclical by themselves or they appear as (parallel) sub-cycles of the informational cycle. And this has to be kept in the concept of the informational cycle!
DF12
[Operators]	:
Eight particular, but at the same time general enough operators of the informational cycle are the operators concerning the cyclical counter-Informing and embedding. It is to understand that these operators have a metameaning and
have to be distinguished from concrete Informings, denoted by capital Latin letters. Particular ire tuope rato rs can be obtained by indexing (specializing) of the general metaoperators. Thus, the following four definitions for these operators can be proposed:

Df

Df
'counter-inforras_in_a_single_
infor!iiational_cycle '	V
' counter-inf orni_in_a_3Ìngle_
informational cycle'	v
' is_counter-informed_in_a_ single_informational_ cycle'	V
'are_counter-inf ormed_in_a_ single_informational_ cycle'
'does_not_counter-inform_in_ a_single_inforraational_ cycle'	V
'do_not_counter-inform_in_ a_single_informational_ cycle'	V
'is_not_counter-informed_in_ a_single_inforraational_ cycle'	V
'are_not_counter-informed_
in_a_s ingle_informati onal_ cycle'
'embeds_in_a_single_
informational_cycle'	v
'embed_in_a_single_
inforraational_cycle'	v
'is_embedded_in_a_single_
informational_cycle'	v
'are_embedded_in_a_single_ informational_cycle'
'does_not_embed_in_a_3Ìngle_
informational_cycle'	v
'do_not_embed_in_a_single_
inforraational_cycle'	V
'is_not_embedded_in_a_single_
inforraational_cycle'	v
'are_not_embedded_in_a_single_ informational_cycle'
■
PX 2
[Operators]' :
Some possible dei'initions of the cyclical operators of Informing are the following:

Df
V <
Df
Df
Df
h E, where C-
^ E, where C
Df
Df
A H^)

In these definitions , > , and H are
CEC	E
cyclical by themselves (by definition). Besides these cyclicities (the informational cycle and within it the cycles of counter-Informing and, embedding) the cycles of counter-Informing and embedding can be understood as informational processes parallel to each other.	■
[Operator] :
Now we nan put the following question: How can the
operator interpreted,
be	informa t ionally
in memory that it was
keeping
determined as h ? It is possible to give the following definitional interpretation:
"Df



This interpretation is a short-form expression of
=Df	-»E- ^C
»	♦	DFIO
The operators f- and -| in [Operators]	are
in some respect counter-cyclical, for |- creates one cycle and the other. These countercyclical processes become obvious in counter-
directional counter-Informings |- and -| and in
C	C
counter-directional embeddings h„ and H . So
E	E
far, we have explained to some extend only the
operators	and -I', According to
DF6
[Operators) , we have tlie following: DF13
[Operators]	:
Let us define the following four cases of complex cyclical operators and explain their meaning :
	^Df	3(1-,	■A).	(h	A	
	"^Df	3(1/,	H).	(1/	A	H)
	"^Df	3(H,		(H	A	1/)
	^Df	3(/|,	h).		A	h)
The short-form notation of these operators are,
/* * /* *
for instance, h , ^ , H , and	,
yj
respectively. The operator H , in fact, is reduced to	for backward cyclical non-
Informing can be simply excluded as an additional explanation. Irrespective of the
relation of dominance, similarly happens to Vj'".
In this way, the operators b'"' and can be reduced into	■
[Operators] :
Two complex cyclical operators of explicit non-Informing can be determined in the following way :
I/'
Df ^Df
3(1/, yi).(|/ A v|) 3(-/|, |/).(y| A 1/)
and
/* /*
These operators can be denoted by I/ and , respectively. Evidently these complex cyclical operators can be reduced to 1/ or -/j.	■
II.1.4. Operators Representing the General Type of Parallel Informing
II.1.4.0. Introduction to Informational Parallelness
Informational parailelness belongs to the most complex and semant ica11y most pretentious problems. The state of the^art of p'arallel Informing will depend substantially on the possibilities hidden in the meaning of the expressiveness of parallel informational ope r a to r s 11 is clear th'at we have to appropriate a sufficiently powerful set of parallel informational operators quite at the beginning of our logical-discourse. This set of parallel operators roust be diversely rich and has to cover imaginable possibilities of parallel informational connectedness, i.e. of parallel communication and informational interaction .
It' became evident already in the framework of the general and cyclical Informing that we have been in many cases confronted with the appearance of the so-called parallel informational processes. If the processes are parallel, there should exist an informational connection among them, via communication and informational interaction. Parts of information are influencing other parts of information or themselves. There can be several levels and forms of information constituting the so-called informational interweave, for instance, on the level of operands and operators. First, we shall show a general (very formalistic) case of parailelness on the level of informational operators, not searching into details of the parallel structure. Then, we shall prepare the philosophy of parallel possibility and necessity of informational interweaving by introduction of particular operators. Next, operators of the informational transfer (communication) wilJ be determined. And at the end of this section we shall deal with informational operators of appearance and vanishing, as well as of choice, memorizing, forgetting, renewing, interrupting, breakdown, and enriching of information.
II,I.4.1. Operators of the General Parallel Informing
In this subsection we shall in fact repeat the philosophy of general Informing, introducing operators of parallel Informing.
DF15 [Operators] :
Two general metaoperators (informational variables) of parallel Informing will be introduced. Let there be the following definition:
¥ V =11
Df
'informs_in_parallel'	v
'inform_in_parallel'	V
'is_informed_in_parallel'	V 'are_informed_in_parallel'
As usual, 11= and =j| are understood to be unary, binary, or multiple prefix, postfix, or infix parallel informational operators. Here the
parailelness concerns the appearance of at least two parallel operators. By introducing parallel operators we are in fact concerned with a system- of parallel informational processes. . The appearance of and means that there will exist at least one more operator of the type |t= and HI. respectively. The operators ||= and =j| point to the connectedness (parailelness) of 'the process possessing |t= or 4 with other informational processes, having operators of the parallel type. The parallel operators-||= and =11 could be particularized and universalized in
DF2
the way we have described in [Operators] . ■ DF16
[Operatora] :
It is evident that similarly to the previous cases of informational operators, parallel operators can be made complex too to obtain their bi-directional function:
11=
NI
Df
Df
3(lt=, =11). (IH A =11) 3(=i|, |N).(m A 11=)
and
The superscript operators in the last two expressions are the so-called feed-back parallel operators. This feed-back parailelness means that in an infix notation the right side process of the expression is additionally parallel connected to the left side (first expression), and vice versa (second expression). The relation of dominance A is understood in the usual manner, so that the left parallel operator dominates .over the right one.
The operators |)= and HI can be understood as
universalizations of the operators lt= and =|1, and ♦ ♦
marked by |t= and . If the universalization of the operators is comprehended as an operational complexion, the particularization always has the meaning of operational specialization. ■
DF17
[Operators] :
DF16
Tn opposition to [Operators]	two general
parallel me taopera tors (informational variables) of parallel non-Informing can be determined:
V HI
"Df
'doea_not_inforra_in_parallel'■	v
'do_not_inform_in_parallel'	v
'is_not_informed_in_parallel'	v 'are_not_informed_in_parallei'
To inform in parallel has the meaning that the process in question informs more than one informational process concurrently. To be not informed in parallel means that at the most one process of Informing has the influence on the process in question.	■
,DF18 Operators] :
In a similar way as in [Operators]	the four
add i t ional , compi ex operators of parallel Informing can be defined,.for instance:
	"Df	3(lt=,	Hl).(lt= A	HI)
	"^Df .		=11). (Ik A	HI)
	/^Df	3(=i|,	Ik). (HI A	Pi
	"^Df	HI'll.	A	IN)
The abbreviated notation of these operators can /♦ + /♦ ♦
be 1^= , It' , =j| , and , respectively. The philosophy of these operators can be quite complex regarding the parallel Informing and parallel non-Informing. It is worth mentioning that distinct parallel operators in the above compositions can be particularized in different ways, so that parallel Informing and parallel non-Informing may not be in the opposition at all.	■
[Operators] :
Now it becomes evident that also universalized operators of explicit parallel non-Informing can be determined. The following two cases can be distinguished as consequently the nonparallel Informing ones:
u4
Df
Df
A\).{\¥ a HI)
and
These operators can be denoted as	and 4 >
respectively. By their appearance these operators exclude the so-called general parallel Informing from informational processes they concern.	■
II.1.4.2. Operators of Parallel Informational Possibility and Necessity
Information and Informing are informationaliy interwoven phenomena. In principle, the interweaving of information is circularly spontaneous and represents merely another form of information, called informational interweaving. This interweaving can be comprehended as informational connectedness or, more generally, as informational parallelness. The question is how this parallelness arises and what are its possibilities and necessities.
'^11' "ll'
DF7
etc.
[Variable]
Let "("ii) mean that information a is parallel
accessible (in an informational input and
output way), so that information is arising
as the possibility of a parallel informational connectedness (communication and interaction) among various informational entities. In this expression Jt^^ is a regular informational
variable of information a. It is to say that n
II
is a regular or an inherent property (an element of informational axiomatic basis) of the arising information a.	■
DF8
[Variables] :
Let	) mean that information oc has a
necessary (concrete) parallel structure v^^ (in
an informational input and output way). Of
course, by v , in a(v ) also the parallel It	II
structure of a is determined, so that a can be decomposed in parallel informing processes which communicate and interact with other informational processes. Besides parallel informational connectedness with other information, v^^ denotes the so-called necessary
parallel structure and organization of a.	■
EXl
[Variables] :
Information of possible and necessary information of parallel connectedness and v^
arises within information a through particular
Informings which can be marked by	and (= ,
It	V
for example. In general, Tt^^ is the information
of possible parallel connectedness. E.g., "n^")
concerns the possibility of parallel connectedness of a. Thus,
A spontaneous way of informational arising does not limit the arising of connectedness among various entities, cases, and parts of information and Informing of information. In static information (algorithms, today's data structures), the connectedness of static informational items is fixed, determined as informational necessity. In general, the interweaving of information can be seen as a possibility and necessity of informational parallelness.
r.r •
[Variables] :
Let us introduce two general markers, variables, and operators tt: and v, denoting the notions of possibility and of necessity, respectively. First, n and v will be used as markers, for instance, marking specific informational operators concerning possibility and necessity. Th.us, it is possible to introduce informational operators
a N t,, (oc) 71 II
denotes the process of arising of information concerning the possibility of parallel connectedness of information ot. Further, (a, ß) concerns the possible parallel connectivity between a and ß. To some extent, Tc^l belongs to an informational entity. So, ß) can belong to a, to ß, or to both of them, if a and ß "cooperate" informationaliy. So	ß) may be split into J^nl")	t||(ß)>
as the possible parallel connectivities of a and ß, respectively.
Similarly, v is the information of necessary parallel connectedness. E.g., '^n^f' concerns the necessity of parallel connectedness of y. Thus ,
t-
A , HI .
7t V II
etc.
Further, it is possible to use Tt and v as variables, for instance in
(*(7C), n(ß), I(v), 7c(C), v(E), Jt, V, etc.
Informational operators, concerning primarily the possibility and necessity, are for example:
T 1=
v„(T)
denotes the process of arising of information concerning the necessity of parallel connectedness of information f• Further, 6) concerns the necessary parallel
connectivity between t and 6. To some extent,
v^l belongs to an informational item. Thus, (t
, 8), for instance, can belong to f> to 8, or
to both of them, if Y and S "cooperate" informationally. It is therefore possible that	S) be split into v^^ ( T ) and ( 8 ) , as
the necessary parallel connectivities of y and 8, respectively.	"
II.1.4.3. Operators of Transfer of Informatiqn among Parallel Processes■
Information of possibility and necessity of informational connectedness is the essential point of informational parallelness. Certainly, informational processes can exist in parallel without any mutual informational influence (relation, transfer, dependence). It means that informational realms of completely isolated processes are separated too.
In this section our intention is to introduce some operators of informational connectedness which realize the so-called proper parallelism of informational processes or briefly the parallel informational system. As we can understand, the notion of parallelness surpasses the so-called sequent i a 1ness , concurrence, and teraporalness of a system and turns philosophically into the structuralness and the organizing principles of information and Informing of information.
,DF20 [Operators]	:
If in an informational formula operators of the types |t= and =j| appear, it'means that this formula possesses some informational operators of connectedness with other formulae. There are several possibilities to determine the so-called operators of informational communication. The aim of these operators is to transfer information from one informational process to the other. In this way, we can introduce the following simple operators of transfer :
T
Df
' c a r r i e s o f r '	V
'carry_off'	V
'is_carried_off '	v 'are carried off
In this definition t is a regular informational operator of carrying-off. As we can understand, this operator can have an active and a passive mode. If a f, information a carries off and if t a, information a is carried off. Further, if afp, information a carries off information ß. It is possible to define t v t instead of
in the last definition.
For the second operator of transfer we take:
Df
•'brings ' 'bring' 'is_brought ' 'arebrought'
In this definition i is a regular informational operator of bringing, which can have an active or a passive mode. If a J,, information a brings and if 4, a, information a is brought. Further, i f a 4- ß r information ct brings information ß or information ß is brought by information c<. It is possible to define V instead of J, in
the last definition.	■
There always exist pairs of particular (fixed) operators t and Further, we can imagine how these operators can be particularized and universalized. We can use )= and [= , for
instance, but also t^
and
4- for marking £
operators of transfer within counter-Informing and embedding, respectively. Universalizations
of these operators could be	and J,^, for
instance. For complexed	operators of informational transfer special symbols can be introduced.
,DF20
,DF21 [Operators] :
According to the definition [operators] , it is possible to determine operators which block the transfer of information. Thus, we can define two informational "blockers" in the following way:
and
4-,
"Df
Df
'blocks_to_carry_off ' •	V
'block_to_carry_off '	v
'is_blocked_to_be_carried_off' V 'are_blocked_to_be_carried_of f'
'blocks_to_bring'	v
'block_to_bring'	V
'is_blocked_to_be_brought'	v 'are_blocked_to_be_brought'
These blocking operators of informational transfer prohibit explicitly the transfer of information in question in those informational systems, which concern the security, for instance, of life information. Further, in the last definitions, it is possible to define ^

and
V	instead of
and

respectively.
EX4
[Operators]
Let it arise information which can be characterized as catastrophic. This sort of information which regularly arises in an informational process can cause catastrophic conaequonces when transferred to another procesa. Let this information be denoted by
Thus, let us have an initial parallel system
where in the information e.
« IN ß, T lt= S
first process catastrophic arises as a consequence of
parallel Informing |h. So,
a IN ß L e

T IK 8
appearance
The operator L marks the nation £
H
block the carrying-off of information e
of
information t^ within information ß. Now, let a
let process
Y block £, to be brought into
H
Thus, there is
the
, and
H
second
(« T^ f.^) IN ß.
(r I,
IN 8
This completes the example using operators of blocking informational transfer.	■
DF2 2 [Operators] :
Now, it is possible to define more general operators of informational connectedness, namely the operators of inter-informa t i ona 1
Informing. These operators inform from the environment of a given information, e.g., from information a, using a particular subinformation £ within a on the way from an informational transmitter to possible informational receivers (sinks). The way between a transmitter and a receiver is performing as regular Informing, so that these general operators of informational connectedness have the function of Informing, similar to the functions of general operators. These operators can certainly be particularized to achieve the functions of operators t and J.. The aim of these operators is to conduct information in an informational system not only through carrying off and bringing of information, but also through regular Informing of this information on the way to informational receivers. In this sense these operators perform as regular operators of Informing. Let it be:
Df
'sends_and_informs'
'send_and_inform' 'is_sent_and_inforraed' 'are sent and informed'
V
V
DF23
[Operators]
We can certainly define two non-inter-informational operators of Informing; these operators can also be understood as general operators of non-Informing particularized in the described manner. Thus, we have the following definitions:
"Df
and
Df
and
'does_not_send_and_does_not_
inform'	V
'do_not_send_and_do_not_inform' V 'is_not_sent_and_is_not_
informed'	v
'are_not_sent_and_are_not_ informed'
'does_not_receive_and_does_not_ inform'
'do_not_receive_and_do_not_ inform'
'is_not_received_and_is_not_
informed' 'are_not_received_and_are_not_ informed'
'Df
'receives_and_informs'	v
' rcce Ì vfi_and__ i n f o I'm '	V
' ia_recei ved _and_j riformed '	V 'are received and informed'
Also in this case, in the upper definitions, operators f ^ and	can be replaced by
disjunctions f v f and 1 v JL .	■
These operators cari be read in the following way:
means means means means
a ir
a. ß t ir a
t a, ß a f ß
a< ß t T.
and
a i
a, ß 4 4 a 4 a, ß oc 4 ß
a, ß Y > S means
means means
m.eans
'a sends and informs' 'a and ß send and inform' 'a is sent and informed' 'a and ß are sent and
informed' 'a sends and informs ß' or 'ß is sent and informed by a'
'a and ß send and inform
T and 8' or 'T and S are sent and informed by a and ß'
a receives and informs' a and ß receive and
inform' a is received and
informed' a and ß are received and
informed' a receives and informs ß' or
ß is received and
informed by a' a and ß receive and
inform y and 8' or f and 8 are received and informed by a and ß'
This completes the explanation of the operatora of inter-Informing t and 4 • In the last two definitions of operators If and 4, it is possible to put t V ■(I' and 4 v 4 instead of
f and 4, respectively.	■
11.1.4.'!. Operators of Informational Appearance and of Informational Vanishing
In this subsection we shall enter into the most substantial discourse of Informing which concerns informational appearance and informational vanishing. Informational appearance is for instance the arising of cowntfi—information and f-he arriving of sensory information to human cortices. Informational vanishing is for instance the dieing of a being's metaphysics or the ceasing of information because of certain changes in an outside phenomenon.
DF24 [Operators] :
Let us have the following definition of the operators of informational appearance:
L V J
Df
'causes_the_appearance_of' 'cause_the_appearance_of' 'comes_into_existence' 'come into existence'
V
V
Information a, which informs, causes the appearance of counter-informa t i on u, or counter-information w is coming into existence through the Informing I of information a. Thus,
a Nj. u) w =lj a
a L 0Ü 0) J a
[Operators] :
It would be very useful for later axiomatic construction that we become aware of the significance of the previous•definition, which directly induces the following axiomatic
formula:
« N
L u)
((L^	A (J^ u))
This formula says the following; if information a informs, then counter-information u is coming into existence. But it is evident that this axiom is only partial, for we can put a more complete one, such as:
( (a N)
(ha))
(L w)
Further, we have some other axiomatic possibilities, for example:
a N a L w

a L (0 a H
and
which yields
a h
etc.
[Operators]
DF25
a L
DF24
According to [Operators ]"""', it is possible to define two operators of informational nonappearance, for instance:
V
'Df
'does_not_cause_the_
appearance_of ' 'do_not_cause_the_
appearance_of ' ' does_not_come_into_
existence' ' do_not_come_into_ existence'
In case of these operators, for instance, the outside information does not appear. It means that the informational environment is not delivering information which could; in the form of, let us say sensory information, arrive to the human cortices. Also, these operators represent Informing which lac)?s the substantial component of Informing, namely the coming of counter-Informati on into existence. In this second case we have to do with a technological system, which does not counter-inform (which does not have the component of informational arising), but can accept i n f b rraa t i on c oni i n g from its environment and produce information in a traditional way, through its algorithmic performing.	■
DF5
[Operators]
EX6
According to [Operators]	it is possible to
construct several axiomatic formulae, as the
DF25
consequences of [Operator]	. Thus, we can
have the following axiomatic presumption:
a ¥
S "
This formula says the following: if information a does not inform, then counter-information u is not coming into existence. However, it is obvious that this axiom expresses only a logical particularity, for one can put the following example:
( (a ti) A (|?i a))
(L^ u)
Further, we can construct another axiomatic possibility, for instance:
Several similar examples can be constructed. ■
The operators of informational appearance indicate also the beginning of a new process, the arriving of new information, and, certainly, the coming of information into existence. In this way, the operators of appearance are specifically involved in the so-called arising of information or in Informing. The operators of informational vanishing are in a certain contradiction with the operators of appearing. They indicate the ending of an informational process, its abolishing. In this way, the operators of vanishing are speci f i (;ally involved in the so-called ceasing or coming of information and its Informing to an end . ■
DF26
[Operators] :
It is possible to have the following definition of the operators of informational vanishing:
T V r
Df
'causes_an_end_of'	V
'cause_an_end_of'	v
'comes_to_an_end'	v 'come to an end'
Information a, which informs, can cause an end of itself (comes to an end) or of information ß. The notations of these facts are:
a 1
r a
and
a n ß
r a
Informational items a, ß...... , which
inform, can cause an end of themselves (come to an end) or, for instance, of informational items	Ti, ... ,	The notations of these
facts are:
a. ß ,
and
Y "I or r a,
a, ß ,
-n,
r
-n. «, ß,
EX 7
[Operators] :
The previous definition can induce several consequences of axiomatic and theoretic nature. The most natural one would probably be
a
a n
where the operator ^ has the meaning of "if
71
'...', then it is possible, that '...'". In this formula we have to do with possible implication or with implication of possibility. If information à informs, then it is possible that its Informing will be stopped (or will come to an end). Further,
a "1
I "1
a
If a is coming to an end,, then its Informing I
a
stops. In this case implication ^ is necessary. Next, it is also possible to express the stopping of a whole process, for example:
(a, ß,
T N T),

14
In this case the particular process vanishes, while its informational components can still perform informationally.	■
DF27
[Operators] :
To the operators of vanishing T and r it is possible to define two opposite operators, the operators of informational non-vanishing. It can be done in the following way:
Df
'does_not_cause_an_
end_of• 'do_not_cause_an_
end_of' 'does_not_come_to_
an_end' 'do_not_corae_to_an_ end '
Through these two operators it is explicitly said that the stopping of an informational process cannot occur in this state of processing. So, the process is forced to continue its Informing. In a deductive theory, for instance, a well-formed formula, which is already a theorem, can be forced to be deductively developed (informed) to become a new theorem. Further; as long as sensors of a living being are intact (or are informing normal ly), sensory information c is appearing continuously (e.g. rhythmically), so a (ff is not coming to an end).	■
[Operators]
EX 8
By means of operators and V^ it is possible
to construct some preaxiomatic formulae. Let us have the following axiomatic presumption:
a N
(a 1
/
a,
This formula says the following: if information a informs, then information a does not cause an end of itself and does not come to an end. The last presumption can be broadened, of course:
((a t=) V (H a ((al
/
A (-1.
/
(r^ a) A
(« r^) )
In the last formula merely the case H « was added to a f=. Now, we can imagine the following implication considering all possible cases of Informing :
oc H ß
( (a a ) A (a
/
1) A
(ß	a) A (ß	ß) A
( (a t= ß)

(a 1= ß)))
This formula presumes that in the process of Informing a |= ß none of its component, including the process of Informing itself, causes an end or is coming to an end.	■
II.1.4.5. Operators of Informational Choice
Informational choice is the inherence of informational spontaneity. Informational choice concerns and presumes the existence and the informational arising of i n f o r m a t i o ri a 1 alternatives, among which in a given sit.iiation none, a single one, or some of them can be
chosen for further consideration of Informing. The choice from a chaos of alternatives can take into consideration the necessity and possibility, a determined and an unforeseeable decision of choice, a simple (predicative) and a complex (informational) selection of informational items, etc.
A mechanism or operator of choice can use its own information of choice, a predicative criterion by which it decides on truth and falsity of a chosen element, an arising and not absolutely predictive mechanism of choice, etc. In our discourse of informational choice we shall use only two typical informational operators, namely, the simple choice (|) and the complex choice (D). Further, it would be possible to introduce operators of choice considering functions of possibility and necessity (for instance, I , D , I , and Q ).
n 7t V	V
DF28 [Operators] :
In this definition, the following simple informational operators of choice without and with predicative criterion will be determined: I ,	and	Let be:
Df
'chooses_among_alternatives'	V
'choose_araong_alternatives'	V
'is_in_the_process_of_choo8Ìng_
as_possible_alternative_(by)' V 'are_in_the_proce3s_of_choo3ing_ as_possible_alternatives_(by)'
The operator | is understood to be oriented from left to right, so that one can use the symbol | , which unambiguously shows the
direction of the operation or	for the
opposite direction. Thus, we get the following meanings :
a I	a chooses among alternatives (which are
not specified yet);
ß r ■ • . , T I
a. ß, ... , T choose among alternatives (which are not specified yet);
I a	a is in the process of choosing as
possible alternative by not yet determined information of criteria;
I a, ß, .... T
a, ß, ... , T are in the process of choosing as possible alternatives;
a I ß, T. ■•• . 8
a chooses among listed alternatives ß, T.....S;
«,ß, ...,Tl?,T,, ...
a, ß, ... , T choose among alternatives ^, r], . . , , ^ or
T), ... , 5 are in the process of choosing as possible alternatives by criteria a, ß.....if.
The criteria of choosing can have the nature of possibility. In this case, the criteria of choice can be chosen. Thus, the operator | has
Jt
the function of choosing possible criteria (from a generative set) and by the chosen criterion to choose alternatives. In contrast to I , the operator | has the fixed, necessary
JT	V
criteria already at a disposal and uses them in a deterministic way.	■
^ DF29 [Operator] :
Let us define the complex informational operator of choice in the forms of Q, 0 i 0 ■ D , and D . Let be:	^
D
'Df
'chooses_among_alternatives_and_
simultaneously_informs ' 'choose_among_alternatives_and_
simultaneously_inform' 'is_in_the_process_of_choosing_ as_possible_alternative__( by_ criteria_...)_and_ i s_simultaneously_informed ' 'are_in_the_process_of_choosing_ as_possib2 e_alternatives_(by_ criteria_...)_and_are_ simultaneously_inforraed'
The consequence of this definition is that the operator □ informs its alternatives by its criteria, and also vice versa (if it is universalized), during the process of choosing. In this way, alternatives depend on the criteria and these depend on the alternatives through the Informing of the choice operator flit is usually understood that the operator Q performs from left to right, where the criteria of choice are on the left and the alternatives on the right of the operator. In this case the
notation Q can be used. The notation D can be -> <used in the case when positions of the criteria and of t he a 1 te mat i ve s are swapped. The criteria of choosing can have the nature of possibility, so that they can be chosen in an inforraationally operative sense. Thus ö has
7X
the meaning of free choosing the criteria in
the case of D > for instance. In the case of n
□ , the set of criteria is fixed by the list of
V
the so-called necessary cho ice-cri teri on information.	. ■
EX9
[Operator] :
Let us explain the meaning of some formulae, using the operators of the type Q. For instance:
a □	a (as criterion) chooses among
.the alternatives (which are not specified yet) and informs;
are informed (and D can be universalized ) ;
a>ß, .-. .rD?. T),.....K
the criteria a, ß, ... , T choose among the listed alternatives t), ... , where the alternatives are informed by the criteria and may also be vice versa.	■
11,1.4,6. Operators of Memorizing, Forgetting, and Renewing of Information
A memory is what is left behind when something happens and does not completely unhappen. This trace does not have to be in a special place, and it does not have to tell much about what has happened.
Edward de Bono [8] 41
In this subsection we shall answer the question "What are memorizing, forgetting, and informational renewing as informational processes, illuminated through introduction of specific informational operators and their operands?" What happens in an informational realm which is being concerned with processes of memorizing, forgetting, and- renewing of information? Obviously, these processes belong to the fundamental and substantial phenomenology of Informing, i,-e. of the so-called informational arising.
DF9
[Variables] :
Let us first discuss the informational realm of entities, items, pieces, and units which constitute the domain of informational operands. Let us introduce the notion of informational lump X, representing a free informational unit with a specific informational structure. Thus, one can imagine that information,, which informs, in fact chooses, generates, changes, and maybe destroys lumps and connects them in a parallel manner as informational entities.
A lump A informs as information. Let the
entities of a lump A be a construct
- ■ * +
A = A , A , A
where
a, p , . . . , -f D
criteria a, ß, ... , T choose among the alternatives (which are not defined yet) and inform;
D a	the alternative a is in the process
of choosing as possible alternative by not yet determined criteria and is informed;
D a, ß, ... , -r
alternatives a, ß, ... , T are in the process of choosing as possible alternatives and are informed;
. a 0 ?, lì . ■ ■ . . ? "
the criterion a chooses among the listed alternatives 75, ... , where a informs and the alternatives
A is the so-called preinformational
(initial or head) part, A is the central (contents or meaning) part, and
+
A is the postinformational (final or tail) part
of a lump. Parts A and A^ connect a lump with other lumps, and vice versa. The preinformational connectedness means that the lump in question has arisen as a consequence (motivation) of Informing on already éxisting lumps. The postinformational ^connectedness of a lump means that this lump represents a kind of semantic basis (motivation), which contributes to the semantics,, of lumps to which it is connected. Certainly, the connectedness of
lumps can be arbitrarily circular and arises spontaneously.
Lumps as informational variables and as given informational entities can be arbitrarily subscripted and particularized. A lump X informs in a parallel way and this fact can be expressed symbolically by the following system:
X =
This system means that during the life of a lump X, it participates in connected processes concerning informational lumps and that it informs also autonomously in itself, according to \ X as a system of cooperating parallel processes
X Ih X X II-- X x"^ IN x'
X Ih X , * *
X Ih X , + *
X Ih A ,
X Ih X , » +
X Ih X , + +
X Ih X
These nine parallel processes constitute the informational lump X as information.	■
[Variables ] " :
The next notion relevant for processes of
memorizing and forgetting of information is the
so-called informational trace. A lump can
participate in several traces. A trace t is an
informat i ona11y connected set of lumps
X , X„.....X , thus
12	n
memorizing net. Memorizing means the appearance of connective information with the arising of thr. contents which subject the connectedness. Forgetting means the \-anishing of ■ connective information with the changing (also arising) of the contents in informational lumps. In gcRfrnl, Informing of information is nothing iiiiir-e than appearing and vanishing of information wit:hin the connected informational entities.	■
In general, memorizi. ng'can be understood to have two components: memorizing as writing (impressing, imprinting) of information to a memory tra<;e (engram information) and memorizing as reading (recalling, remembering) of impressed information from a memory trace. Ttic corresponding informational operators of impi-essing and recalling will be subscripted by m_w (memory_writ ing) and ra_r (meraory_reading), respec v. i \-e] y .
I Operators I	:
M em <5 ri z ing means informational mapping or
Ftiforming of informatior\ cr into a trace	-r.
MeinorizLr\g is a Itind of embedding	of
information into another information, is	an
operation of informational impressing	or writing and can be represented by the operator
IN V =11
m w m w
Df
'is_impressed_in'	v
'are_impressed_in'	v
'is_an_impression_of'	v ' are_.impressions_of '
The connectedness of lumps in a trace is
conditioned by their head and tail parts, +
X and X , respective 1 v. Memorizing means to i i + proceed through the X -part of a lump to the X -part of another (next) lump to which the previous lump is informationally connected. Such a proceeding is possible also in the opposite direction, from the X -part of a lump to the X -part of a previous lump. The connectedness of the tail and the head parts is bi-directional, so, in the processes of memorizing Informing can be performed in both directions. This structure of the trace is an abstract construction. There could be many head and tail parts in a trace for its backward and forward connections to other traces.
The trace as information informs in itself. This Informing can be described by the parallel system
The informational operators |h	and 4
m_w	m_w
represent the adequate parallel Informings concerning the impressing of one or several informational entitles in a trace or traces.
lOperators] :
From the previous definition of the memory writing operator we have, for instance:
f IN	information o- is impressed in (a
m_w
trace not yet identified);
a,. 0-^, ...
rn
m w
t ho. informational ... , :t
enti ties
m
1' "2-
are impressed in (traces not
yet identified);
the trace t is an impression of (information not yet identified);
" "2.....\
t ih t
IN T, ,
m w I
'2'
DF 1 1
(Variables]
A memorizing net is a parallel informational system of informational traces. Thus, for a net V we have:
v = t^, t^,
0- h=
V Ih V
the informational traces t^, t^, ... ,
T uri! impressions of (informational n
entitles not yet identified);
information <7 is impressed in the trace t or the trace t is an impression of information £7 ;
Memorizing and forgetting is n unique process of informational af>poarMni:(; ami vanishing in a
a 11=	T
m m w
operators' part icularI za t i on is used. We have the following definition:
the informational entities cr , a ,
... , cr are impressed in the truce t m
or the trace t is an impression of
informational entities tr , ct , ... ,
1 (d
cr' ; in this case the informational
entities <r,, <r„, ... , ct are
12	m
condensed as impressions in the trace
t; we can say that the informational
entities cr , c , ... , <t are
12	m
condensible memorized in the trace t;
11=	T ,
m w 1
V HI m I' ni r
nr
'recaiIs'	v
'recall'	V
'is_recalled_from'	v 'are recalled from'
The informational operators lb	and 4
m_r	m_r
represent the adequate parallel Informings concerning the recalling (reading) of information impressed in memory traces.	■
■	EXll
(Operators] :
Let us looit at the most general example of recalling information by means of the operator
.=11
information cr is impressed in the
traces t
1 '
n
or the traces
T are impressions of
r 2'
information a; in th i -s case information c is distributed ]. y memorized in the informational traces
1
IN -r
m m w 1
the informational entities cr ^ , cr^ ,
... , a are impressed in the traces m
T , T , ... , T ; in this .case, the
12	n
informational entities <7^, cr^, ... ,
<7 are distributed as 'impressions over m
the traces r . t„, ...' , t ,
12	n
irrespective of the values of m and n;
we can say also that the informational
entities ct , a , ... , tr are
12	m
distributively memorized in the
informational traces t t , ... , t .
1 2	n
This example completes the possibilities of
using the memory writing informational
operator |t= . For the informational operator m_w
HI we can show similar formulae.	■
m_w
Information which is memorized in informational traces can be recalled (reminded, remembered, recollectedY from informational traces by means of particular informational operators. However, it, is to stress that during memorizing of information in informational traces this information is subjected to the Informing of traces and to the Informing of arriving information which concerns information memorized in traces. It means that during the memorization of' information in traces this is changed and is regularly ari s ing -wi th in informational processes in traces. This changing and arising of information during its memorization in traces can be called the informational forgetting.
[Operators] :
Information, memorized in a trace or several traces, can be recalled (recollected) by means of particular informational operators. Recalling of information means its reading from memory traces, so the subscript ra_r for
CT =il T m m w 1
the informational entities <t , o- ,
1 .Z
... , cr are rec;alled from the memory ni
traces r , t , ... , t or the memory 12	n
trnr'(;s t.
recall the
informational entities cr^, ct^» ••• >
tr ; in this case information impressed m
in the memory traces is delivered in the form of informational entities
Forgetting of impressed information can be understood as nothing else but Informing of memory' traces by similar, outward information, and Informing in the sense of informational decay of memory traces by themselves. The consequence, of such Informings is the changing and arising of originally impressed information.
DF32
[Operators] :
Let us define the following two informational operators of forgetting:
V HI,
Df
' disintegrates_and_
informs'	V
'disintegrate_and_inform' V 'i s_disintegrated_and_is_
informed'	V
'are_disintegrated_and_ are informed'
Informational forgetting is conditioned by disintegration (decomposing, losing, forgetting gradually) and Informing (by new incoming information) of impressed information. If incoming information informs impressed information, this kind of Informing represents a form of impressing of incoming information. ■
[Operators] :
As we have defined, the process of forgetting is influenced by two components: the decaying of impressed information by itself and the memorizing (impressing) of incoming information. In this respect forgetting is a parallel informational process, for instance,
« Ih^. -r
18
This process can be decomposed into the system
forgetting, and renewing of information.
t	t, a ih^ t
or into a simiiar parallel informational system
II. 1.4.7. Operators of In for/national Interrupt and Breakdown

a it.	t
m w
where the o pern tora of H i a i n to R r i» I. i o n (decaying) and impressing are used. The i n forniti t ionu J, variable t rtìpresenLs u iiiemox'y trace or a net of such traces.	■
The notion we have to introduce next is the informational similarity. Memorizing and forgetting as informational processes hide informational similarity in their backgrounds. Similarity of information r«^present.s one of the basic relations, similarly to informa t i ona] difference. To be similar means to be different to some extent and different means to be similar to some extent.
DF33
[Operators] :
Let us have the following definition:
Df
'informs_similarly'	v
'inform_similarly'	v
'is_informed_simìlarly'	v 'are_informed_simi lariy'
As usually, this operator (relation) can be bi-directional, for instance	or	for
Informing from the left to the right or vice versa, respectively. The operator of similarity can be understood to be symmetric, so, for example,
a ß = ß a
Similarity is a substantial condition of memorizing, forgetting, and renewing of information.	■
,DF34 [Operators]	:
Renewing of information belongs to the so-
called cyclical parallel Informing. Its general
informational operators could be discussed in
Section II. 1.5. Renewing of information informs
in cycles composed of simultaneous impressing,
recalling, and forgetting. In this process new
information is entering (informing) memory
traces. Thus, the operator of renewing |f= can
rnw
be defined by the parallel informational system in the following way:
ah T rnw
Df
Cf IN. T, T IN
C ,
IN T.
m_w
T, CT IN J. T
In this system was replaced by |N . . In a
similar way it is possible to define the
operator =(| . In the process of informational rnw
renewing t has the intention to recons.truct some information which contributed to the formation of the trace t. In this process there can be additional influence from sensory information o- which currently enters into the informational system in question.	■
In this subsection we have not been dealing with operators of the types II;; and particularizing them for impressing, recalling,
At a certain point of the proccss of Informing, it is possible In say thai, the current Informing of (original.) information was i n l.c r riipl.od through tlie so-called interrupting information and that after interrupting a new, the so-culled brealtpoint information takes ooritfol OVIM- thf process of Informing within the original information. If the new information is typically counter-informational, that is opposite and unexpected in respect to thc^ original, 1 nten-upted information, then the now 1 n f o trn i Mg is cal Ind breakdown. A breakdown as information informs in the opposite, essentially new sense, however still considering the interrupted information. The interrupting information, which causes interrupting of the governing information, can be counter-informational or outside-informational .
In our dlsco\irse we have identified three informational entities and their Informings: the original (interrupted), the interrupting, and the breakdown information. In a process of Informing, the phenomenon of interrupting original information and then Informing by the so-called breakdown information is called the informational turn. This turn brings into the Informing of information a new informational circularity and spontaneity which regenerates and renews previous (original) information. The informational turn is one of the essential events in the context of counter-information and counter-Informing. However, the turn is not only a new circularity; through spontaneity it is also a new informational quality, a new kind of information. The condition for the informational turn is the appearance- of an interrupt and a breakdown. Let us now define some operators of these types of Informing.
[Operators]	:
Let us have the following definition for informational operators of interrupting:

"Df
'interrupts'	v
'interrupt'	v
'is_interrupted '	V 'are_interrupted'
Further, let the operators of informational breakdown be defined as follows:
V II bdw Ijdw
Df
'breaks down'	V
'break_doMn'	v
'is_broken_down'	v 'are broken down'
These operators constitute the possibility of examination of the so-called phenomenology of informational turn.	■
EX13 [Operators] :
Let n be the original, n the interrupting, and ß the breakdown information. For the sake of this example, let [j be the so-called metaphysics, i.e. the informational entirety of a living being. Let us have the following
parallel informational system:
n IN n.
(a being's contemplation) (interruption of contemplation)
<ß breaks down |.i)
ß, n |t= ß, M-> ■ (ß and ti inform mutually)
ß 11= n	(however, ß informs dominantly)
A
In the third line of the above system C means
£
that ß is informationally embedded (C) by e in p: and in n (interrupting information).	■
II.1.4.8. Operators of Informational Enriching
The enriching of information does not mean merely the accumulation of various information but also, and above all., the informa t ionaJ connectedness of the existing information. Rich information is characteristically interwoven and informs in an informationally interactive way. A scientific discipline, for instance, can be understood as information which through its sclentLfic arising enriches itself. The arising of informational connectedness means the arising of new informational relations, forming new connectedness, which delivers additional scientific understanding. In the process of informational enriching, informational quantity and informational quality arise. Under informational quality also the arising of a new meaning and diversity of meanings can be understood.
Consequently, the informational enriching concerns the arising of quality and quantity of information in parallel (concurrently, simultaneously, and in an interweaving form). The informational quantity alone may not necessarily contribute to an essential enriching of information. Parallel and parallel cyclical Informings maybe the most natural ways of informational enriching. It is of course possible to conceptualize various particular and universal operators of informational enriching, with the aim to study general and special methods and theories of informational enriching and its informational arising. Operators of enriching can cause the arising of measures and criteria of informational quantity and informational quality of information and informational processes.
II.1.5. Operators of Parallel Cyclical Informing
In this section we shall join and develop the concepts exposed in Sections II. 1.3 (Operators^ Concerning the Informational Cycle) and II.1.1 (Operators Representing the General Type of Parallel Informing). In this section, the philosophy of information which has been developed, comes to questions such as the following: Is parallel Informing among different infot;mational cycles at all possible? Are informational cycles conceptualized in principle and in n concealed manner as time-dependent sequences or can this time-
dogmutical, causal, or i f - th on-princi p 1 ed glimpse be surpassed informationally? Is it possible to view a sequence of informational cycles as a complex informational process, which hides informational parallelness among different informational cycles? Does this concept exceed the possibilities of informational processing in a living organism and in a technological machine because of the so-called time barrier?
In this section we shall not block our discourse and the development of a theory of parallel Informing by the use of any time-dogmatical barrier. ;Thus, we shall allow parallel Informing also among different informational cycles. We could say that time in this discourse will function as any other regular information, which is, in its nature, causal, consecutive, sequential, serial, etc. We shall not accept a concept of an informational cycle as time-dependent constT-uction' among cycles in an informational process, but as informational process in its entirety and its informational temporality.
DK36
[Operators] :
Two general operators (operational variables) for expressing the parallelness of the informational cycle can be introduced by the following definition:
It- V HI
Df
'informs_in_parallel_in_a_
single_inf ormational_cycle' 'inform_ in_parallel_in_a_
si ngle informational_cycle' ' i s_i n formed_in_parallel_in_a_ single_inf ormational_cycle' 'are_inf a rmed_i n_parallel_in_a_ single_informational_cycle'
This, definition means that a process within an informational cycle informs into other processes of its own cycle and into the processes of other informational cycles, and is informed from the jjrocesses of its own and from other informational cycles. It is to understand that operators |l- and HI can be particularized and universalized according to the needs of their application. Further, it is to comprehend that similarly to Section II. 1.4 (particularized) parallel informational operators of non-1nforming, universai ization, possibility, necessity, transfer, appearance, vanishing, choice, memorizing, forgetting, renewing, interrupting, breakdown, and enriching of information can be constructed. ■
DF37
[Operators] :
Let us define the usual universalization of the parallel cyclical operators in the previous definition as
II-
Hl
HI
Df
"Df
3(|h, HI) . (Ih A HI)
3 ( HI, Ih ) . ( HI A Ih )
and
In the first expression, the operator HI is the so-called parallel, cyclical feed-back operator (which is subordinated to the mairi operator |h); so is Ih in the second expression (which is subordinated to the main operator HI). We can again introduce
20
Df
DF38
and
Df
[Operators]
As a contrary to the operators ||- and HI, two parallel cyclical metaoperators (informational variables) of parallel cyclical non-Informing can be introduced by the following definition:
II/ V 4
'Df
'does_not_inform_in_a_
parallei-cyclical_way' 'do_not_inform_in_a_
parallei-eye1ical_way' 'is_not_informed_in_a_
parallei-cyclical_way' 'are_not_informed_in_a_ parallei-cyclical_way'
If Informing is not aimultaneously parallel and cyclical, it can still be generally parallel or generally cyclical. The operators \\/ and HI can, for instance, hinder the parallelness in cyclical Informing, cyclicity in parallel Informing, or both of these types of Informing within general Informing.	■
DF39
[Operators]	:
As in the definitions of general, cyclical, and
DF6
parallel types of operators ([Operators] ,
DPI 3	DP 18
[Operators]	, and [Operators]	,
respectively, it is possible to define the following four cases of complex parallel cyclical operators and to explain their meaning:
Ih^'	^Df	3(11-,	HI) .	(Ih	A	HI)
	"^Df	3(11/,	HI).	(II/	A	HI)
Hi'^	^Df	3 (HI,	11/).	(HI	A	11/)
Hl'^	^Df	3(HI,	Ih).	(HI	A	Ih)
The short-form notation of these operators can
be 11- , IV I HI , and HI , respectively. The operators in complex definitions can be particularized non-uniformly, for instance, as parallel cyclical counter-Informing and parallel cyclical. (Miibcdding. The next sl.op of \in i versa! i t i on or romp 1 oxa t. i on would lie l.o complex the already complexed opei-utors.	■
DF40 [Operators]	:
Two complex informational operators of a strict parallel cyclical non-Informing can be defined in the following manner;
11/
HI
Df
Df
3(|k, Hi), di' A HI) 3(HI, IH) . (HI A II/)
and
These operators can be marked by |!y and HI , respectively. Obviously, these complex parallel cyclical operators can be reduced to II/ and HI- ■
DPI 1
[Operators]	:
According to [Opera tors ]^ , parallel cyclical
operators of counter-Informing and embedding
can be defined. It is to understand that If- and
u
IF^ are parallel within an informational cycle.
We can list the following four definitions concerning the informational cycle:
^ ^'c
nf

Df

Df
counter-inforras_cyclically_
in_pnrallei '	v
counter-i nform_cyclically_
i n_parallei'	v
i s_oounter-informed_
cyclical1y_in_parallel' V are_counter-informed_ i:yc 1 i ca 11 y_in_parallel '
does_r\ot_coun ter - inf orm_
0	yr1 i c a1 1y _ i n _pa r a11e1' v do_not_counter-inform_
oyclioally_in_parallel' v i s_not_counter-informed_
eye Iica1ly_in_parallel' V are _not COun ter-in f orraed_ cyclioaI1y_in_parallel'
ombeds_cyclical 1 y_in_
parallel'	v
embed_cyclically_in_
parallel'	V
i s_embedded_cyclxcally_in_
parallel'	v
are_embedded_cyclically_in_ paral lei'
does_not_embed_cyclically_
ln_parallel'	v
do_not_embed_cyclically_
1	n_parallel '	v is not_ nmboddcd. eye 1 i cal 1 y_
i n_paral lei'	V
are_not_erabedded_
eyeli cn1ly _ in_parallei'

[8]	E. de Bono: The Nechnnicm:^ of Mind. Penguin Books, 11 a r m o n d a IV o (■ t h , Middlesex, England (Reprinted 1977).
[9]	G.W.F. Hegel: The Science of Logic. In Hegel's Logic, being Part One of the Philosophical Sciences (1830), translated by W. Wallace. Oxford, At the Clarendon Press, reprinted 1985.

Df
TEMPORAL LOGIC - A TOOL FOR SYSTEM MODELING
INFORMATICA 4/1988
UDK 519.711.3
Bogdan Dugonik Technical Faculty Maribor
Abstract
This paper presents a model for processes, that cominunicate by message-passing, and the use o-f temporal logic -for their description. The specification of process determines behaviour on its external ports, whence its internal behaviour is not significant. Specifications based on the model are compositional, that is, we get external behaviour of a system from specifications of its components. In the introduction a process recorded with traces is presented. A model is shown that represents a process specification by use of temporal logic. Temporal logic operators, examples of temporal assertions and some specifications for primitive and composed processes are given.
Povzetek
V članku Je prikazan model za procese in mreže procesov, ki komunicirajo izklJuCno s prehajanjem sporoCil, ' in uporaba časovne logike za njihov opis. Specifikacija procesa določa obnašanje na njegovih zunanjih priključkih, dogajanje v njegovi notranjosti pa nas ne zanima. Opisi procesov so komponibilni, kar pomeni, da zunanje obnaäanje mreie dobimo iz specifikacij njenih komponent. Pri tem prikri Jemo njeno notranjo strukturo. V uvodu predstavimo zapis procesa s sledmi. Prikažemo tudi specifikacijo procesa in mreže procesov z uporabo časovne logike. Opisani so operatorji časovne logike, podanih Je nekaj primerov časovnih izjav ter specifikacij enostavnih in sestavljenih procesov.
Introduction
This paper treats processes that cominunlcate exclusively through message passing. Messages are transmitted over input and output ports. Let us observe a process on its external ports for a time, up to some moment. We get a sequence of events. The sequence gives us a trace. The trace Is an information about behaviour of the process up to that moment. Traces are appropriate to describe only terminating processes. The model of a system is modular, it Is composed of some primitive processes. In a system output ports of processes are linked to inputs of another processes. Some input and output ports are not linked. Nonlinked ports are external ports and linked ports are Internal ports.
A process is determined by a ' set of possible input-output behaviours . The behaviour on a set of input and output ports is an infinite sequence of observations ;
6 = (t»,Ina,,Out«,,Rda.) , (ti > Ini ,Outi ,Rdi ) , . . .
t is a trace of events on input ports (input events) and events on output ports (output events) .
Events represent communication between processes. Communication can be synchronous or asynchronous. If communication is synchronous, a process cannot send a message until the receiving process is ready to read the message on his input ports. If communication is asynchronous, a process can send message on his output ports as soon as it is ready without having to wait for the receiving process. In case of synchronous communication, input events present the data read by a process, and in case of asynchronous communication, they present the data that have appeared on input ports. Output events in both ways of communication present messages sent by a process at its output ports. An event is recorded in the trace as a pair (x,k), where x Is a datum and k is the port name where the event has appeared. In and Out are boolean functions. If they are true for input - output ports, then we say the process is ready to
communicate. If logical values of some input port functions ace true, then we say the process is ready to read on those ports. If logical values of some output port functions are true, then we say the process is ready to send on those output ports . Rd gives us the number of messages read up to the observed moment for each input port . It is clear that this number cannot be greater than the number of messages which have appeared on an input port. We could describe the process by writing down all possible sequences 6, which is vague and for more complicated cases practically impossible. Because of that , we apply a language which enables us to express all possible behaviours of a process.
Temporal logici
Natural languages are very expressive, but not precise, whereas formal languages are very precise but not so expressive. Formal languages have strictly defined semantics and syntax. It is impossible to say everything in them. But what you can say is unambiguous.
Temporal logic is a formal language. It allows us to make a temporal description of a process and also formal dealing of it. He present a model of temporal logic with infinite sequence of states
3it), Si , Sa,
Assertions in temporal logic have a temporal meaning because they include some temporal operators. Language employs a set of basic symbols: individual variables and constants, function and predicate symbols. These are divided into two subsets i a set of local symbols and a set of global symbols. Global symbols keep their values unchanged from state to state. Local symbols can vary their values from one state to another. We use boolean operators, classical operators and temporal operators. Existence quantification and universal quantification are also included in the system.
The operators of temporal loqict
By w we denote a term or a formula that may contain some temporal operators.
-	temporal operator "always":
□ w means; w is true in this moment and will be true forever.
-	temporal operator "eventually":
O w means: w will be true at least in one of following moments.
-	temporal operator "until":
Hi "« wa means : Wi is continually true until Wa, becomes true, and wa does indeed become true.
Temporal aasertlonai
Formulas given below are examples of some temporal assertions. They are compounded by using temporal operators mentioned above. We will write A for conjunction, -i for negation,
and =!> for implication.
□(u ❖ O V) - whenever u is true, it will eventually be followed by v.
O D w	- eventually w will become true
and it will remain thrue fotevett
Q O w	- every instant is followed by
some later instant, when w is true, so w is true indefinitely often.
O V <>((-1 v)H u) - if v ever happens, its first occurrence is preceded by, or coincides with u.
We choose some fixed reference moment named reference moment when interpreting a temporal assertion. It tells us, what holds at the reference moment, i.e. now, and what will happen in the future of it. It says nothing about what happened before. Instead of 6 « Um, Si,S3t,..	let us write 6 =
where 6i is an observation in behaviour <S . An observation can be treated as a state. Usually, a state of a process reflects current values of variables of the process. Hence, an observation Si is a state, which involves the history of process up to a current moment in its trace ti . Specification of a process signifies how this process behaves. For this model we write process specifications in the form of temporal assertions.
The specification of the system P has the forni <P> R , where R is an assertion, written in temporal logic. Specification <P> R is reads each behaviour of system P satisfies assertion R. <P> R is external specification, if R specifies behaviour on P's external ports. System specification is a conjunction of component	assertions.	The	external
specification is obtained by using a proof system.
The proof system for specifications Includes rules for temporal logic, rules regarding the description of the domain of values, rules for designing a system, a set of system components with their precise specifications, and axioms which define the behaviour of the system. Specification defines two kinds of process properties: safety and liveness. Informally, safety specification asserts, what the process may do, and liveness specification asserts, what it must do.We will give some examples of specifications.
First, we show how certain ports of a process can be disabled, discarded from communication by using negation operator -i with In or Out function.
In the specification below input port Si and output port bi are disabled. Input al can not read the data, and bl can not send the data.
-	temporal operator "unlesa":
Hl M Ha means: wi is continually true or there is a moment, when wa is true and up to the moment wi is continually true.
-	temporal operator "next":
o means: w will be true in next moment.
< C > D -1 In(ai )
Out(bi)
Now, let us assume that communication is asynchronous. We will write external specifications of processes. That is, we will specify a process only by means of lengths of ports' traces, without using In, Out and Rd. By Ibi we denote the length of trace on port b. bEc means, that sequence b is a prefix of
sequence c. First example below (fig.l.) gives a process with input port k and output port 1. The process reads six values on input k and writes them on output 1 leaving out the first two values.
Qj	fl		
		1	c 1
bi i	B	d r—	
			
fig.l.
Process port 1.
with input port k and output
< A > O 1 E [ k(2),k(3),k(4),k(5) ] A ( 111 = 0 Jf |k| Si 3) A (O |k| = 3 =i> O 111 = i) A (O |k| =4	O 111 = 2)
A (O |k| = 5	O 111 = 3)
A (O |k| 6 =t> O 111 = 4)
The first line is safety specification. It indicates that at iiiost four values may occur on the output port, which will be read on input port k. The second conjunct indicates that the length of the output trace is zero, unless the length of input trace is three. The third conjunct and those following require the length of the output trace to be increased by one as soon the length of the input trace is increased.
Second example (fig. 2) is a process with two inputs and one output. It reads one value on j and one value on k. Then it writes first the value on the output 1 from input k, and later the value from input J.
Fig.2.
A process named B with two inputs one output
and
< B > O 1 E [ k(0),J(0) ]
A ( O |k| = 1 =t> O 111 = 1 ) A ( 0( IJI = 1 A |k| ^ 1) =i> D 111 = 2)
Finally, we present an asynchronous network P of component processes A,B and C. Components A and B have one input and one output, component C two input^s and one output. Components are linked as shown in figure 3. Process A reads one value on input a and writes it twice on output c. Process B reads one value on input b and writes it once on output d. Process C reads values on its inputs, then writes them on output e as follöwsi first a value from input d, then both values from input c.
Fig.3. Process P with visible and hidden internal structure.
First, specifications for each components are given :
<A> acC[a(0),a(0)]
A ( O lai 1 O □ Id =2)
<B> DdECbCB)]
A. ( O lb! = 1 O □ Id) = 1)
< C >	D e E [ d(0),c(0),c(l) ]
A	( O Id I 1 =t O le I = 1 )
A	( O ( Id = 1 A Idi ^ 1) ❖ ( O lei = 2)
A	( O ( Id = 2 A Idi ìi 1) ( O lei =3)
External specification of P is obtained from conjunction of component specifications by using proof rules.
< P > D e E [ b(0), a(0), a(0) ] A ( O Ibi Sì 1 ={> O lei = 1 ) A ( O ( lai à 1 A Ibi à 1) =t> O □ lei = 3 )
It tells us how the system behaves on its external ports.
Conclugion
We presented a model for temporal logic as one languages for specifying temporal logic is its a temporal	behaviour
Specifications can be sho that they need no extr arise, if we want to because in the model we only qualitatively events.
processes which uses of existing formal them. The benefit of bility of describing of	processes,
rt and so expressive, comments. Troubles set the exact time, can express ourselves bout the time of
RefTencest
[1]	Dugonlk B., Modeliranje procesov, diplomsko delo, Tehnifika fakulteta v Mariboru, 1987
[2]	Hoare C.A.R., Communicating Sequential Processes , Prentice - Hall International, Ltd., London 1985.
[3]	Kapus T., ćasovna logika in mreže procesov, diplomsko delo, Tehniéka fakulteta v Mariboru, 1987
[4]	Lamport L., What Good is Temporal Logic ?, Information Processing (1983),pp. 667-568
[5]	Manna Z.,Pnueli A., Verification of Concurrent Programs, Part I» The temporal Framework; Tehnical Report STAN-SC-81-836, Standford University, June 1981.
[6]	Nguyen V., Demers A., Gries D., Owicki S., A model and temporal proof system for networks of processes. Distributed Computing (1986), 1, pp. 7-25.
COMPACT IMPLEMENTATION OF THE PROGRAMMING LANGUAGE PROLOG IN THE ENVIRONMENT OF Me POPLOG
UDK 681.3.06(PROLOG:POPLOG]
Ivan Bruha Me Master University Department of Computer Science and Systems Hamilton, Canada, L8S 4K1
ABSTRACT
There nrc uiany sofiwiire tiiskw Uiii.t coiikl be solved more efficiently if they used both proceduraJ and declarative programming techniques'. One practical and useful solution is to combine the programming languages Prolog and POP-11 into one environment called POPLOG.
This paper discusses the wiiys of constructing a Prolog compiler in terms of POP-11 data structures and control techniques. Afterwards, the McMaster version of the Prolog compiler and the implementation of a few built-in Prolog procedures are introduced; moreover, the communication means between POP-11 and Prolog in Mc POPLOG (the McMaster-version of POPLOG) are described.
1. INTRODUCTION
There exist many software engineering problems that utilize both procedural ajid declarntive prograuuning techniques. Several attempts to combine the programming language Prolog [3], [9] with procedural programming langua.ges hn.ve been done such as:
-	Prolog and LISP, e.g. LOGLISP [13], QLOG [10],
-	Prolog and POP-11, i.e. POPLOG [11],
-	Prolog and Modubi,, [12].
We are convinced that the combination of POP-11 and Prolog is a very useful and adequate solution. Although both LISP and POP-11 [2] are AI programming langun.ges, the programming language POP-11 provides more data structures and control terlniiquos that can bo ustati for inq>l(^ni(^nt:i.tiou of Prolog terms and its inference nui.chiu(;. The lirst version of such "a system that combines the la.nguii.ges POP-11 and Prok)g has been proposed and implemented at Sussex University, U.K. This multi-language environment is called POPLOG [11].
Although the idea of joining two high-level AT progra.m-ming languages into one system (POPLOG) is excellent,' the actual Sussex implementation can be improved in .several n,spcct.s. Therefore the author of this paper has decided to devel(.)p and implement a completely new version of POPLOG (Mc POPLOG) [5], [6] with the.se attributes:
-	the entire McMaster version is written in C (rather than in POP-11),
-	the system is very compact, since the Prolog subsystem utilizes only standard POP-11 data structures and control mechanisms without any change,
-	it has straightforward couimnnica.tion means between both languages,
-	the POP-11 subset has been substantially simplified by returning partly to the Edinburgh's POP-2 [7], taking new features of Sus.scx's POP-11 [2] that are important to AI project.s (not to system tasks),
-	the McMaster version is much shorter and faster (see Chapter C for a, (ximparison), and i.s portable to any 32-bit machine.
This paper describes the way t)f de^'cloping a Prolog compiler in terms of POP-11 data, structures and control techniques. Chapter 2 explains POP-11 structures necessary for such an implementation (see also [2], [7]). Chapters 3.1 and 3.2 introduce the fimdainental model of a Prolog compiler written in POP-11 (details are in [11]). Chapters 3.3, 4, and 5 describe the McMaster model of a Prolog compiler, in POP-11, the implementation of a few Prolog standard procedures, and the communication means between POP-11 and Prolog in the Mc POPLOG. Chapter C compares all three models (the fundamental,'Sussex, and the McMaster ones).
2. CLOSURES AND NON-STANDARD CONTROL STRUCTURES OF POP-11
The programming language POP-11 [2] is a descendant of PÜP-2, a language originally developed at Edinburgh University for AI research in 1971 [7]. Now, there exist a few dialects of POP-2; the most famous are POP-11 (for PDP-11, VAX-11, and others) and 'VVonderPOP (for DEC-10). POP-11 is mostly used for list processing, like LISP, but it uses Pascal-like (or Algol-like) well-structured syntax, has a, large set of various data structures including records, vectors, strings, arrays, references, pairs, lists, dyna.mic list.'i, closures etc., and involves many non-standard control stnic-tuics such as function-unwinding, processes, backtracking, ])a.ttern matching, da.ta.ba.se jHoccssing, macro facility, conipiling-diuing-execution facility, and so on. Like BASIC, it is fully interactive, dialogue la.ngua,ge. On the other hand, POP-11 has Ijeen mostly implemented'as a compiler.
Ill this chapter, W(; introduce only closures and nonstandard control mcrhanisMis of POP-11 that are necessary for explaining how a. Prolog compiler is written in POP-11. Other necessary syntax and built-in functions of POP-11 are in the Appendix. See also [2], [4], [7].
If a function, say f (a,b,c,d) is called ofien for some values c = cO , d = dO , we can create a, new function g(a,b) so that
g(a,b) = I(a,b,c0,d0)
for all values a , b . This function is called a closiirc . A closure ■ is defined in POP-11 so that the original function is followed by a sequence of actual arguments enclosed into decorated round parentheses ('/. and '/,) , i.e.
i a co.do ■/■) -> g;
The ai'guments cO , do are called frozen values, the function f is called a frozen fxmction, and the entire process is called partial application.
One common use of the technique of closures is to save (or freeze) values of some varia.bles at a certain time for a later call, if the var iables are expected to change their values. Tliis feature is especially utilized by a so-called continuation backtracking, discussed later.
The programming language POP-11 offers a lai'ge collection of non-standard control structures that can be used to construct efficient and elegant solutions of problems which would be inefficient to solve using convenient control structures. The Prolog compiler written in POP-11 needs the following control techniques: function-unwinding and backtracking.
1. Function-unwinding. If we want to quit a function and immediately call another function, the POP-11 function chain can be u.sed. Tlio function rail chain(G) invoked within a function F unwinds the call of the fmiction F and iiiuiiediately calls the function G .
function alpha(x);
betaCx);
end;
function beta(ic) ;
/♦here after executing function gamma ♦/ /* called by chain(gamma) within beta »/
chain(gamma) ; Z+unuinds the call of beta and calls »/ ...	/* immediately the function gamma */
end;
function gamma; /»called by function-unwinding mechanism»/
end;
If we call the function alpha , the function beta is invoked in a standard way but chain (gamma) causes the call of the function beta to be unwound, and the function gamma is performed. After executing the function gamma the control returns to the function alpha after the i)oint where beta wjis invoked.
The function chain is us(xl for the tail-<)i)tiuiÌ7.ation of Prolog procedures in POPLOG.
The function-iniwinding mechanism of POP-11 cnn be applied not only for a single function call, but more than one call can be unwound. Here we need only these functions:
-	the function chainto(F,G) unwinds all function calls up to (but excluding) the function F ; from here the function g is call(!<l in a standard way;
-	exitfrom(F) unwinds all function calls up to imd including the function F ; from here the execution continues in a standard way.
2. Continuation backtracking. The programming language POP-11 involves two types of backtracking that can be u.sed for many AI applications. The first type, state-saving backtracking.
has special keywords and constructs for saving nodes of a search tree. It is quite simple, but needs a large memory. See e.g. [1],
[4], [7].
The other tyi)e, continuation backtracking, does not have special constructs, l)ut only a Sfx^cial interpretation of function calls. In ordinary programming, a function call can be seen as a rec(uest to reach a goal. When a goal has been successfully performed, control returns to the im'oker where'the request was done, and continues. In the continuation backtracking, a function call is still seen as a request to achieve a goal. The difference is that the called function is told what should be done next if the goal is completed. This specification of what is to be done next is passed as an extra argxunent to the function, and is called continuation. Thus, the continuation backtracking has this interpretation;
-	if the goiil succeeds the function does not return to its in-voker, but the continuation is applied;
-	if the goal fails, the function returns to its invoker; thus normal return from a function is interpreted as a failure.
Example. Consider the following puzzle [1]. Each positive integer n (which is not divisible by 3) can be constructed by multiplying by 2, or adding 3, starting from 1, e.g.
10 = ((( 1 * 2 ) * 2 ) -)- 3 ) -f 3
The function alg(n) solv(\s this puzzle liy contin\iation back-tra.cking:
function alg(n); vars list;
algl(lambda; rev(list); exitfrom(alg) end, /»initial continuation»/
1, nil)
end;
/»here nl is the current number, initially 1 , list involves a sequence of actions, initially nil »/
function algl(contin,nl,list); if nl == n then
contino	/»success -so continue »/
endif;
if nl < n then
/»firstly try goal »2 »/ alglCcontin, nl»2, 2::"»"::list);
/»fail: try +3 »/ algKcohtin, nl+3, 3::"+": .-list)
/»if nl>n then return: failure */
end;
alg(lO) =>
»» [»2»2+3+3]
If a. goal comprise.s two .subgoals then in ordinary programming, wi- <-.all Ijoth subgoal.s in a .sc(|iicii<-.(^ E.g. let a goal (function) g(x) have subgoals a(x) and b(x) . Then in ordinary programming:
function g(x); a(x); b(x);
end;
/»call the 1.subgoal»/ /»then call the 2.subgoal»/
In continuation liacktracking, the suljg<ial b will be passed as a continuation ai-gument' to the fvuiction a :
function g(contin,x);
a( bC/. contin, x Z), x)
So, if we invoke g wè have to supply explicit continuation as to what is to be done when the goal g is successfully accomplished. The function g invokes the subgoal (function) a with the continuation
b(y. contin, x 7.)
Consequently, if the function a succeeds, the above continuation will be invoked, i.e., the function b will be invoked with the originell continuation and tlic fro/,<;u value of it .
Thus, a sequence of goals in the continuation backtracking is realized by the closure incchanisiii. No other constructs arc necessary. The continuation liacktracking is obviously faster (2 to 4 times, task-dependent) than the state-saving backtracking and does not need so much memory. Therefore, this type of backtracking is the basis for the Prolog implementation in POPIL
dereference them, i.e., look at the contents of the innermost reference. For that purpose, we use the auxiliary function darel(X).
3.2. Fundamental model with explicit re-uninstaiitiation
The unification algorithm for pattern matching is one of the most important mechanisms of Prolog. To demonstrate fundamental ideas of compiling Prolog procedures, the P0P-1Ì fxmc-tion unify on Fig. 1 can be introduced as a unification algorithm (see [11] for details). It assumes Prolog constants aie POP-11 numbors or words, Prolog vaiiablcs are POP-11 references (with recognizer iaref, and accessor cont), and Prolog structures are POP-11 pairs (with recognizer ispair, and acces-sors front, back). The last statement with the unify call tries to unify front(X) and front(Y) , and if it succeeds, backCX) and back(Y) are unified. If even they match, the given continuation contin is invoked.
3. A PROLOG COMPILER WRITTEN IN POP-11
3.1. Prolog variables
Any implementation of Prolog must solve many problems, among them the most important ones seem to Ijc:
-	unification,
-	backtracking,
-	instantiation of Prolog variables,
-	the 'cut' operator.
Evidently, the only type of liacktriw.king suitable for Prolog im-plenieritation in terms of PüP-11 is the coutinu.itioii backtracking. This technique is utilized without any change. The 'cut' operator cćin be implemented by utilizing the POP-11 function-unwinding techniques. In this chapter, we survey how the remaining problems can be solved. Other interesting ideas of Prolog implementations can be found in [8].
Unfortimately, a Prolog variable cannot be directly represented by a POP-11 variable, because we could not express the matching of two uninstčmtiated Prolog vaiiables in such a representation. Moreover, theré would also arise problems with the backtracking. The only adequate representation of a Prolog variable seems to be a POP-11 reference; it is a stćmdard record of one component (see Appendix).
(1)	An uninstantiated variable X is represented by a reference containing the word undef . This is done by the POP-11 function consrel :
consref(undel) ->X;
i.e., a reference with contents undef i.s assigned to x .
(2)	If an uninstantiated variable X matches a term T , then the value of T is attached to the vMiable x . In POP-11, this is done by the accessor cont that can select or update contents of any reference:
T ->cont(X);
(3)	If two uninstantiated vaiiables X and y share then one reference becomes the contents of the other, e.g.
y ->cont(X);
i.e. the Prolog variable (POP-11 reference) X contivins a pointer to the Prolog variable y and both of them are uninstantiated. This chaining of shared uninstantiated variables can be done to £iny depth. Therefore, if we process Prolog variables, we should
The above model of unification undoes any changes it has made when it retm-ns to its invoker, i.e. when a fail takes place, see the statements for re-uninstćmtiation:
undef ->cont{X) , undef ->cont(Y)
However, the representation and matching of a general struc-
tiue f(al,a2.....an) with fvmctor f , arity n , and components
al,a2,... ,an is unclear. One way to represent internally stich a structure is as a list
[ f al a2
an ]
but it would be very cumbersome and time-consuming. Note the actual models do not use such an internal representation of Prolog structures.
If we accept the above function unify as a fundamental model of unification, then Prolog clauses can be easily compiled to POP-11 functions. In the following, we discuss how Prolog fa<;ts and rules with conjunctions and disjunctions of goals are compiled. See [11] for details.
A Prolog fact is compiled by means of the function unify. If there are more facts of the same predicate in the Prolog database the return-fail mechanism of the continuation backtracking is simply utilized. E.g. let the database contain
girl(Silva). girl(paulina).
then the predicate girl is compiled as the following POP-11 function:
function girl(contin,X);
/»l.girl: "silya" is POP-11 Bord»/ unifyCcontin, X, "silva");
/♦fail: try another girl»/ unifyCcontin, X, "paulina")
end;
A conjunction of two or more goals in a body of a rule is compiled by means of closures as continuations. Consider a conjunction of two goals:
likes(X) :- girl(X), drinks(X).
then the second goal will be placed as a continuation for the first goal, i.e. the above procedure is compiled as
function likesCcontin.X);
girl( drinksC/.contin, X'/.), X)
function ijnify(contin,X,Y) ;
deref(X) ->X; deref{Y) ->Y; if X == Y then
contino elseif isref(X) then Y ->cont(X); contino ; undef ->cont(X) elseif isref(Y) then X ->cont(Y); contino ; undef ->cont(Y) elseif is|)air(X) and iS|jair(Y). then
7* X , Y structures of the same type*/ unify( unjfy(% contin, back(X), l)ack(Y) t), front(X), front(Y) ) endif	/*otherwise fail (no 'else' here) */
end ;
/»unify constants*/
/* X is uninstantiated var*/
/* X instantiated to Y */
/*success: continue*/
/*fail: X re-uninstantiated var*/
/* Y is uninstantiated var*/
/*fail: Y re-uninstantiated var*/
Fi((. 1. Function unify .
The function girl is called to unify X with a girl in the database. Assume Prolog database contains two girls siiva and Paulina . The function girl succeeds with X = silva . Afi-erwards, the function drinks is caJled as a continuation within the function girl . If the second goal succeeds, i.e. silva drinks, then the given continuation contin is invoked. If the girl Silva does not drink (more precisely, the system docs not have an information about her habits), then the function drinks returns to its invoker, and ajiother girl is processed. If no girl drinks then the call of girl returns to likes , and likes fails immediately.
If more than two goals are in a conjunction in a (■lau.s<^ body then closures ns continuations must be nested. E.g.
likes(X) girl(X), drinks(X), slim(X).
is compiled as
function likes(contin,X);
girl( drinksC/. sliitiC/. contin, X'/.), XV.). X)
end;
A disjunction of goals, or a sequence of clauses for one procedure is compiled by utilizing the reiurn-fail mechanism of the continuation backtracking. E.g. l(;t the proceduro drinksCX) be defined by these two clau.se.s;
drinks(X) :- wine(X). drinks(X) :- beer(X).
then the covr<ispoiidiiig POP-11 fmictiou is function drinks(contin,X);
wineCcontin, X); beerCcontin, X);
/• X drinks wine»/ /»fail: try beer */
end;
Example. Prolog proceduro inember(X,L) defined as
momb0r(X,[XI ] ) . inember(X,[JT]) :- member(X.T).
can be compiled as the following POP-11 function, using only the function unify :
function member(contin,X,L); vars T;
/♦1.clause: 2.arg unified with [Xl_3 */ unifyCcontin, L, X::consref(undef));
/» : : is constructor for pairs»/
/»1.clause fails: try the 2.clause»/ consref(undef) ->T;
/»2.arg unified with [.ITD ♦/ unifyCmember('/,contin,X,T'/.) , L, consref (undef ); :T)
end;
We ca.li see that
-	the way of compiling Prolog procedures is very simple, s t raightfor ward,
-	the result of compilation, i.e. a POP-11 function is slow in execution,
-	the principal di.sadvant.ago of this model is that structures are constructed many times only for the purposes of the head matching, c^.g.
X :: consref (undef) for matching [XIJ consref (undef ) ;: T	[.IT]
Th<^ Su.sscx- vci sioii of POPLOG u.scs il very detailed compilation of head matching (sec [11], pp. 161-1G2) so that no structures are constructed for the purposes of head matching. It represents precisely the idea of the continuation backtracking and compiled Prolog procedures as POP-11 functions are fast in execution. A c(3mparison of all models is introduced in Chapter C.
3.3. Auxiliary stack of instantiated Prolog variables
M<; POPLOG d<«.'K not u.se the detailed way of compilation but rather utilizes a, set of a.uxiliary functions that recognize a 'tyi»;' of a j>roi'cssr<l iiiguineut (i.e. whether the argument is a constant, uninstantiated variable, or a structure). Furthermore, the mechanism of instantiating Prolog variables is using an auxiliary .sta<-k, rather tha,u a.ssigmnent .statements for explicit re-uninstiuitiation.
Mi'Ma.-il.i'i- iinxlcl |fil ii.sc.-i iMi juixiliary .stark for storing in-.stantiated Prolog variables (so-called Prolog variable stack). The current offset to tlie Prolog variable stack is designated varoff. If a Prolog variiililc is instantiated then it is pushed onto the Prolog variable stack. Secondly, when n branching point in a search tree has been reached, the system saves the current offset of the Prolog variable stack in the variable varoff 1 .
If a fail occurs, then the system returns to the branching point and rc-uniiistautiatcs all Prolog variables that have
function punify(X,Y); vars i; derefCX) ->X; (leref{Y) ->Y; if X == Y then true
elseif isref{X) then
Y	-> cont(X);
X ->varstdck(varoff); varoff + 1 ->varoff; true
elseif isref(Y) tlien X -> cont{Y);
Y	->varstack(vdroff) ; varoff + 1 ->varoff; true
/♦identical terms: result is true */
/* X is uninstantiated var*/ /*assiijn term Y to var X */ /*fjush X onto the Froloy var stack*/
/*result is true: matchimj succeeded*/ /* Y is uninstantiated var*/ . /*a5Siyn term X to var Y */ /*|)iish Y onto the Proloii var stack*/
/*result is true: iiiiitchin() succeeded*/ elseif ispair(X) and is|)air(Y) then
/*both terms are pairs*/ punify(front(X) ,front(Y) ) and punify{back{ X) ,liack(Y) ) elseif isvector{X) and isvector(Y)
and datalenoth(X) == ddtalenyth(Y) then
/*hoth term are general structures of*/ /* the same arity*/ fur i from 1 to dataienijth(X) do
ifnot punify(X{i ) ,Y{i ) ) ttien /»unify i-th components*/
false.return ^ end if enddu;
true	/*functors ä all components match*/
el se
false	/*otherv/ise result is false */
endi f
end;
function passign(X,T); T ->cont(X); X ->varstack(varoff); varoff + 1 ->varoff
end ;
/*assiqn term T to Prolo<i var X */ /*push X onto the Prolog var stack*/
H'l;]. Kunctions punify and oassiiin .
function meinber{contin,X,L); vars T, varoffl; varoff ->varoffl;
deref(X) ->X; deref(L) ->L;
if isref(L) then
passignU. X: iconsref (undef ) ) elseif ispair(L) then
ifnot punify(X,front(L)) then
yoto 'clausei? endif else goto clause:^ . endif; contini); c-lauseif:
retrieveO;
if isref(L) then
consref(undef) -> T; passiyn(L, consref(undef)::T) elseif ispair(L) then
l)ack(L).-> T else goto clause3 endif;
chain(contin,X,T,Rienber) clause!!: end ;
/*local variables*/
/*save current offset to the */
/* I'rolog variable stack*/
/*1.clause: head matching*/ /* L is uninstantiated var*/ /* [X I _] is assigned to L */ /* L is a pair*/ /* L = [X I J ? */
/*no budv of the 1.clause- so continue*/
/*fail: 2.clause*/
/*uninstantiate Prolog vars*/
/*head matching*/
/* L is uninstantiated var*/
/*nev/ Prolog var to T */
/* [_ 1 T] is assigned to L */
/* L is a pair*/
/* L = [ 1 T] */
/♦body optimized: (|uit tins call and */ /*call immediately member(contin,X,T) */
l-ig. 3. Procedure iiiember compiled by llcllaster Prolog coi'ipiler.
been instantiated iifèer the l5r<i.nching point has l)een encountered. The re-uninstantia.tion is done by the auxiHaiy function retrieveO that uninstantiates all Prolog variables in the Prolog variable stack Isetwecn the currcut offset varolf and the initial offset varoffl :
function retrieve:
while varolf > varolfl do varoff - 1 varoff; undef -> contCvarstack(varoff)) enddo
end;
Note the Prolog variable stack is iniplcinciitixl as a POP-11 vector, and varstack is the bottom of this stack.
The unification algorithm of Mc POPLOG is written as an ordinary function, without the continuation mechanism. The function punily(X,Y) (s('(> Fig. 2) rctui'iis true iff tin; tfU'nis X and Y match, and instantiates Prolog variables if necessary. Note tha.t a Prolog structure I(al,a2,a3) is represented as the POP-11 vector { 1 al a2 a3}.
Besides the a.l)ovo vuiificiition function, few other func.tions can be invoked within compiled Prolog jiredicates. Among them, the function passign(x,T) assigns the term T to the Prolog variable X : see Fig. 2.
Example. The procedure memberCX.L) is compiled in Mc-Master Prolog as follows: see Fig. .3.
The McMaster model of a Prolog compiler
-	follows the idea of the continuation backtracking,
-	constructs no structures in the head of any clause,
-	Prolog structures are represented more efficiently as POP-11 vectors,
-	compiled Prolog procedures as POP-11 functions are fast in execution, and not so large,
-	the system does not need so large system stack (for storing function calls) since msuiy auxiliary functions of this model are written as convenient functions without the continuation backtracking,
-	Warren's tail optimization is always done.
Note that in the actual implementation, the functions deref, isref, and ispair used above are joined into one auxiliary function so that the execution is even faster.
A comparison of thi; all three models i.s dcjue in Chapter G.
also quite simj^le. Among them. Fig. 5 introduces and repeat. The function papply(contin,X) used here considers the terra X as a goal, and applies it with the given continuation.
The 'cut' operator cut-Kcallerd))
is considered as the goal
where callerCl) is the invoker of the current function, i.e. the p.ucnt goal of tl»^ goal tlint <'ontains ' ! ' in its body; e.g. if
g a, b, c. b :- d. e, !, f.
then the aljove 'cut'	oj^cnitor is considered as cut.l(g). The
function cut.l calls	chainto :
function cut.l(contin,F); chainto(F,contin)
end;
5. COMMUNICATION MEANS BETWEEN POP-11 AND PROLOG
5.1. Top level
The Mc POPLOG allows the user to call POP-11 functions from Prolog and vice versa. This iiilcr-language commimication can bo done to any d<;pth.
When the user is a.t the top k-vel of the system POPLOG then the prompt : is displayed and the system expects any POP-11 statement or any Prolog question. If a text begins with ?- then this ?- is recognized as the operator of a Prolog question, otherwise it is considered as a POP-11 statement. Therefore, typing in ?- followed by a Prolog question will activate the Prolog sub.systcui. E.g.
member(2,[l 2 3]) =>
: ?- read(X),write(X).nl.
I maria.
maria
X = maria
yes
/•POP-11 statement: is 2 a member of the list [12 3] ? ♦/ /»answer of POP-11 subsystem»/
/»Prolog question»/ /»term read by read(X) »/ /»printed by Prolog's Brite(X)»/ /»instantiated variable displayed»/ /»answer of Prolog subsystem*/
/»prompt of POPLOG is waiting for next statement or question»/
4. COMPACT IMPLEMENTATION OF BUILT-IN PROLOG PROCEDURES
The Prolog sub.system of the Mc POPLOG uses the above model with the Prolog Viuialile stack, th(! functions punify(x,Y), passign(X,Y), deref(T) (^tc., and Prokig va.viabk^s n^pic^sentcd as POP-11 references. Wo will now describe how built-in Prolog procedures are implemented in this inod(;l.
First of all, the unification operator = iuvok(;s dii'ectly the function punify; its POP-11 representation is on Fig. 4. The operator is uses the function punify, and the function popval(P) that evaluates the Prolog term P as a POP-11 expression (see also Chapter 5.2); its POP-11 representation is on Fig. 4.
1
Fig. 4 involves some other built-in procedures that can be written in terms of the operator is, or they utilize the functions punify or popval (see also Chapter 5.2 for details).
The operators for constructing goals and backtracking are
Prolog clauses can be asserted into the Prolog database at the top level using the standard procedure consult(user) or [user] . POP-11 statements and Prolog questions and clauses can be also used together in a file.
5.2. Calling POP-n fmicUoiis from Prolog
The user of Mc POPLOG can call any POP-11 function from within Prolog. I'^ir tha.t purpose, s(-uianl.ics of the standard Prolog operator is has been oxtende<l within the system POPLOG, and two additional unary procedures popval and popvalfail have been supplied.
popval(P) evaluates the Prolog term P as a POP-11 expression. The expression P must not return any result. This goal succeeds only once.
popvalfail(P) evaluates the Prolog term P as a POP-11 expression; one result must be returned, and if it is POP-ll's false it is interpreted as a fail of the given goal. This goal
function e()_2(contin,X,Y) ; if pun1f.y(X,Y) then
contini) end i f
end ;
/* X = Y */
function is_2(contin,X,Y);
if |)unify(X,popval(Y)) tlien
contin() end if
end ;
/♦ X is r :- X = ()ui3va){Y) . */
function lt_2(contin,X,Y) ;
if popval(X) < popval(Y) then
contino endi f
end;
/* X- < Y ♦/
function var l(contin,X);
if isref(dL'ref(X)) then
contin{) end 1 f
end;
/* Vdr(X)	*/
/* popvalfai l(isref(closure(X))) */ /*if X is uninstantiated var*/
function functor_3(cuntin,T,K,Ar) ; /* fijnctor(T ,h',Ar ) *J vars i;
deref(T) ->T; deref(F) ->F; deref{Ar) ->Ar; if isnunil)er(T) then	/* T is fl niiiiilier*/
punify(T,F) .and punify(Ar,U) elseif isref(T) then	/* T is uninstantiated var*/
punify(T, {IF,	/* T is unified with F(_._, ... ,_) */
for i from 1 to Ar do ' consref(undef)
L'nddo %])	/»I'rolag structures are PDI'-ll vectors*/
else	/* T IS nonvar */
punify(F, T(l)) and	/* F- is unified with functor of T */
puiiify(Ar, datirieniitli(T)-l) /* Ar 1s unified witli arity of T */
end if ;
If /*result of punify*/ then /*if natchinq has succeeded continue*/
contino endif
end;
^
Fly. 4. Implementati on'of some built-in Proloi) procedures
function or_2(cont1n,X,Y); vars varoffl;
varoff -> varoffl;
papply(contin,X); retrieve!);
end;
chain(contin,Y,papply)
/* X ; Y */
/*save the current offset to Proloi) */
/* var stack»/
/*cal1 the l.tioal*/ '
/*fa11: uninstantiate Prolog vars*/
/*try the ü.poal with the tail */
/* optimization; quit this call an<t */
/*cflll imedlately pap()ly(cont 1 n,Y) »/
function repeat_ü{cont1n); vars varoffl;
varoff -> varoffl; repl:
contino ; retrieve!); (joto repl ;
end;
/* repeat */
/♦save the current offset to Prolog */ /* var stack*/
/*fa11: uninstantiate Prolocj vars*/ /*l00()*/
Fig. 5. Pl)P-ll representation of
and repeat
32
succeeds only once.
X is P evaluates the Prolog term P as a POP-11 expression and unifies the result with x ; P must retiu-n just one result.
The Prolog term P in the above procedures ( popval, pop-valfail, is) is comprehensibly written in Prolog syntax l;>eca.use the Prolog conipilcr reads this text. How(m;r, it is <w;ijuate<i a,s a POP-11 cx])ression. Hero are the rules for POP-11 c.valuaiiov. of Prnloc; tcnns (the rules (iii), (iv) will be explained later):
(i)	if P is a constant (or an instantiated Prolog variable) then the constant (vahic of th(! Prolog vnria.l>l(;) is push(;d onto the POP-11 user stack;
(ii)	if P is an uninstantiated Prolog variable then its PÜP-11 representation (a reference) is pushed onto the POP-11 user stack (see the rcprcsenta,tion of Prcjlog terms as POP-11 items);
(iii)	closure(P) where P is a Prolog term is evaluated by pushing P onto the POP-11 user stack;
(iv)	V -> A where V is an instantiated Prolog variable, A is Prolog atom, is evaluated as follows: the value attached to the Prolog -vnrialilo V is nssigned to POP-11 variable A ;
(v)	the Prolog strnc.turc f (PI.....Pn) where f is n-ary functor
and PI to Pn are Prolog terms, is evaluated as follows: firstly PI to Pn are <-vnluated as POP-11 expressions and then the POP-11 function f is called.
Thus, the POP-11 avalnaiion of a. Prolog term P matches the standard evaluation of POP-11 expressions in the progr;unming language POP-11 [2].
Note the POP-11 c.valuaiiov. of a Prolog term. P does not affect the POP-11 user sl,a.ck hccausc;
-	popval(P) must not n^turn nny result,
-	P of popvalfail(P) returns just one result that is popped from the stack, and its value decides a.bout success/failure of the goal,
-	in case of X is P the value of P is popped from the stack and unified with X.
If POP-11 tvdualion of a PtoIoij ir.r oes not r<;l.urn the required nurnlier fif results, a. run-time error occurs.
Example. Define the Prolog procedure concat(Ll,L2,L3) (L3 is a. concatenation of the lists Ll, L2) for the instantiated Li and L2. We will use the standard POP-11 fnnc.tion join for list concatenation:
evaluated as POP-11 ______expression
Fiq. 6. The body of the procedure concat (listSare expressed as dot-pairs)
after matching
concat(L1,L2,L3) L3 is join(Ll,L2). Now
?- concat([1.23, [a,b], R).
R = [1, 2, a, b]
yes
However, the question ?- concat([l,2], [aa(l),bb], R) .
would cause an error .since the entire argument join( [J ,2] , [aa(l) ,bb3) would be consid(H(-(l as a POP-11 exi)r(Äsion so that a non-existing POP-11 function aa would be called with the argument 1 , see Fig. C. To prevent a subexpression of an argument of the procedures popval, popvalfail, is to be considered as a, POP-11 expression one has to r.ln.tr. it \i,sing the fimctor closure as follows:
concat(Ll,L2,L3) :- L3 is join(closure(Ll), closure(L2)).
evaluated as POP-11 expression
l_cl_os_ure___cTosijre_ J
/\ 2 []
aa

/\
bb []
Fig. 7. The arguments of the function are 'closed' now
join
Now Ll, L2 are considered lis Prolog t.ernis rather than POP-11 expressions, .see rule (iii). Thus
?- concat([1,2], [aa(l),bb], R).
R = Cl, 2,- aa(l), bb]
yes
will return iwpected r<-siiM.s, see l''ig. 7.
Exami>lc. Consider Prokjg procedure member(X,L) : X is an element of the list L . If the procedure is used only for instantiated X and L, then - in some cn.ses - it could be defined in (.(^rins of the. st;indard POP-] 1 function member :
member(X,L) :- pop»alfail(member(closure(X),clo3ure(L))).
To [irocess a value of a, POP-11 variable within a Prolog ])rograiu the standard POP-11 function valol (ralue of a variable) has to be used:
: vars list;	/»declare a POP-ll včiriable»/
: [10 9 a ab aaa 0] ->list; /»assign the given list to it*/
/»sort list by Prolog's sort »/
If we wrote
?- List is list, sort(List,S).
the Prolog variable List would be instantiated to the atom list. To introduce a value of l.he POP-11 variable list, we nmst do:
: ?- List is valol(list), sort(List,S), write(S), nl.
[0, 9, 10, a, aaa, ab]	/»printed by arite(S), nl »/
List = [10, 9, a, ab, aaa,	0] /»instantiated variables »/
S = [0, 9, 10, a, aaa, ab]	/» displayed»/ yes
The cvaliui.l.ii)ii of valof is (^iiilxxlicd in tlu^ mie (v).
A POP-11 expression in popval, popvalfail, is ean also contain an assignment to a POP-11 variable, using the POP-11 operator -> , see rvile (iv).
Example.
vars list, sorted;
[10 9 a ab aaa 0] ->list; /«sort list by Prolog's sort »/
?- List is valof(list), sortCList,S), popval(S ->sorted).
/»result of sort assigned to ♦/ /» POP-il variable sorted »/
List = ClO, 9, a,^b, aaa, 0] S = [0, 9, 10, a, aaa, ab] yes
; sorted =>
** [ 0 9 10 a aaa ab]
/»POP-11 variable sorted is »/ /» the sorted list»/
Tlius, thanks to valof and -> we can use POP-11 variables as inpuU to Prolog progra.ms, as well as (mipiiU for storing results.
5.3. Prolog called Ijy POP-11 luiictioiis
Similarly, any POP-11 function can call a. Prolog program by introducing a Prolog qviestion in its body. When a Prolog question is executed within a POP-11 function
-	the Prolog subsystem does not wait for user's replies;
-	but returns true ( false ) as a re.sult, if the Prolog question has succeeded (has failed);
all Prolog variables in the Prolog question will retain their values when retiuning from the Prolog c|ue.stion to the POP-11 function so that they can be processed as any other POP-11 vjvriable.
Exanlpl(^. Definì' POP-11 function sort(list) using the standard Prolog procedure sort(Ll,L2) :
function sortClist) => R;
■/» R is explicit result of function»/ ?- List is valof(list), sortCList,R).
/» R retains its value from Prolog »/ /» question*/
end;
Now, the POP-11 expression sort([10 9 a ab aaa 0]) yields [ 0 9 10 a aaa ab]
Example. Define POP-ll function inember2Cx,l) using the Prolog procedure member(X,L) defined as
member(X,[XIJ).
member(X,[.IT]) :- member(X,T).
We utilize directly a result returned by the Prolog question;
function raember2(x,l);
?- X is valof(x), L is valof(1), member(X,L). end ;
If the goal member(X,L) succeeds then true is returned, otherwise false . Thus true / false becomes the result of the POP-11 function member2 .
6. CONCLUSION
It is important to emphasize that in order to economize time and memory requirements of the system, the entire Prolog suljsystem of Mc POPLOG has been written in C rather than in POP-11. We have introduced the ideas in terms of POP-11 functions for better and easier readability.
Secondly, a Prolog compiler could translate Prolog clauses to POP-11 functions, as we have .shown abo^'e, and afterwards call the POP-11 comi>il(;r which would translate a text of a POP-ll functicjn to a machine code. In the actual implementation, the Prolog compiler translates any Prolog procedure directly to the machine code. All the above examples of compiling Prolog procedures are written in POP-ll to simplify the explanation of principal ideas.
It wi)s not a straightforw:jrd luocedure to compare the a.b(wc tlnce modi.'ls (the fundamental, Sussex, and McMaster one) of Prolog compilers using the POP-ll structures, since
-	the fundamental model (Chapter .3.2 and [11]) is not a part of any POPLOG version,
-	we had at disposal the executable (binary) file of the Sussex POPLOG and [11] only (not a source file, nor any detailed Information ).
Therefore:
-	we have emulated the fundamental model in POP-ll (McMaster version),
-	we have compiled the procedure member on the emulator, Sussex version, a,nd McMaster version,
-	we have run the question
?- memberdOO, [1,2,3,
,100]).
The entire comparison is in Fig. S. As the timings depend on the percentage of calls of standard and user-defined procedured, we have added the execution time of the following question, as well:
?- functor(T,aa,100).
The results presented in Fig. 8 Hiatch quite well a long term statistical (;xperiments we have done for the comparison of both actual implementations of POPLOG; see Fig. 9.
We characterize the McMaster version of POPLOG as a compact one since the Prolog subset utilizes only standard POP-ll data structures, as well as POP-ll control mechanisms, without any change. Thanks to that approach, the Mc POPLOG is vmbelieva.bly shorter, and also faster.
The communication means between POP-ll and Prolog in the Mc POPLOG are quite straightforward and simple. Neither programming language has been changed so that a user unfamiliar with POP-ll can use the Prolog subset only, and vice versa. Only three new predicates (popval, popvalfail, closure) have been added to the. Prolop; subset, and the si^mantics of Pro-l<.)g's is has be(;n slightly extended.
The Mc POPLOG ruiis currently on VAX systems and on systems based on the Motorola 6S000 processor. Detailed information on the system can be obtained from the author.
emulator of the fundamental i:iO(Jel
compilation time
# of machine instructions for member
execution tine of the i)uesticn ?- member ...
execution time of the question ?- functor ...
Zl
127
Sussex HOHLÜG
3a
60*
35
42
Mc
PUPLÜÜ 4
3y IS 12
Fig. 8. Comparison of the Prolog compiler models written in POP-11
Compilation and execution time is in cs (centiseconds), i.e. Iiundredths of seconds
The emulator of the fundamental model has been tested in Mcllaster version of PÜPLÜG
* Estimate done according to the example in [11], pp. 161-2
McMaster Sussex
Comparison: 9 times shorter
3 times faster 2 times faster
3.7 times faster 2.7 times faster
(1)	the system needs memory
(2)	PUP-11:
-	compilation time
-	execution time
(3)	Prolocj:
-	compilation time
-	execution time
75 KU
670 KU
Fig. y. tiemory requirements and timings of the actual versions of POPLOG
REFERENCES
[1]
Anderson, B.: Programming languages for AI: the role of nondeterminism. School of AI, Rep. 25, Univ. of Edinburgh, 1972
[2]	Barrett,R., Ramsay,A., Sloman,A.: POP-11 - a practical language for artificial intelligence. Ellis Horwood, New York, 1985
[3]	Bratko,I.: Prolog - programming language for artificial intelligence. Addison-Wesley, 1986
[4]	Bruha, I.: User's guide to POP-11 of McMaster POPLOG. Dept. Computer Science and Systems, McMaster Univ., Techn. report 87-04, 1987
[5]	Bruha, L: Reference manual of McMaster POPLOG. Dept. Computer Science and Systems, McMaster Univ., Techn. report 87-03, 1987
[6]	Bruha, L: Compact implementation of Prolog as a part of the environment of McMaster POPLOG. Dept. Computer Science and Systems, McMćister Univ., Techn. report 8705, 1987
[7]	Burstall,R., Collins,D., Popplestone,R.:. Programming in POP-2. Edinburgh University Press, 1971
[8]	Campbell,J.A. (editor): Implementations of Prolog. Ellis Horwood, New York, 1984
[9] Clocksin,W.F., Mellish,C.S.: Programming in Prolog. Springer-Verlag Berlin, 1984
[10]	Komorowski, H.J.: QLOG - The programming environment for Prolog in LISP. In: Clark, K.L. and Tarnlund, S.: Logic Programming, Academic Press, 1982
[11]	Mellish,C., Hardy,S.: Integrating Prolog in the POPLOG environment. In: [8].
[12]	Muller, C.: Modula - Prolog: A software development tool. IEEE Software, pp. 39-45, Nov. 1986
[13]	Robinson, J.A. and Sibert, E.E.: LOGLISP: An alternative to Prolog. In: Machine Intelligence 10, Ellis Horwood, 1982
Appendix
To understand the POP-11 programs introduced here we should know only:
vars X. y;
declaration of POP-11 variables y e ->i;
value of the expression e' is assigned to the variable x
value of the expression e is printed and preceded by ♦»
function f(x,y);	end;
declaration of the function t with foiniiils y lambda (x,y); ... end;
declaration of a no-iiamo function (lambda-expression) with formals X, y
iinot C then ...
the same as if not{c) then ...
delimiters of a POP-11 word; thus alpha is a variaLle wlicre<i.s "alpha" is a word (atom)
isnumberC*)
returns true iff the it(!m x is a. number consref(x)
constructs a reference with contents x contCr)
function that selects or updates the contents of the reference r isreiCx)
returns true iff the item x is a reference
front(p) / back(p)
function that selects or updates the first/ second component of the pair p
ispair(x)
returns true iff the item x is a pair nieniber(x,xl)
returns true iff x is an element of the list xl rev(xl)
returns a list of the elements of the list xl in the reverse order xl : : x2
operation that creates a pair from xl and x2 join(xl.yl)
returns the concatenation of the lists xl ajid yl { ••• }
brackets for a vector construct s(i)
the i-th component of the structure s datalength(s)
returns the number of components of the structure s
ON DELETION OF REFLEX ELEMENTS FROM CIRCULAR DOUBLY-LINKED LISTS
UDK 519.683
Mirjana Stojanović, Ivan Stojmenović Institute of Mathematics, University of Novi Sad
ABSTRACT
An element x of a doubly-linked Girellar list is called reflex if f (pred (x) ,x ,succ (x))
We Dropose an algorithm to delete reflex elementsfrcm the list and apply algorithm to give the correct code of Graham 's algorithm for determining the convex hull of a finite set of points in the plane. We show that the code given by Preparata, Lee and Shamos was incorrect and give some corrections to the results of Koplo-witz and Jouppi.
Key words and phrases: convex hull, algorithm, circular list CR 1.3.5 Conputational geonetry and object modeling
AMS Mathanatical subject classification(1980): Primary 86U05, Secondary 68010.
1. Deletion of reflex elements from circular doubly-linked lists
Suppose we are given a circular doubly-linked list S and a function f (pred (x ) ,x , succ (x )) the value of which for given element x of S depends on its predecessor pred(x) and successor succ(x) as well. The element x is called reflex if f(pred(x),x,succ(x)) ^0. The following algorithm will delete all reflex elements from the input list S, starting at a given initial element START. The result is stored in the list CH{S), initially equal to S. Although in general the result depends on the choice of START, in applications such as one rriven in the next section the output is the same for any choice of initial element.
BEGIN
CH(S) = S; h:=false; g t=false; V:=START; r:=succ(r) ; REPEAT
IF f(v,succ(v),succ(v)))> 0
THEN BEGIN {successful test} IF r=succ (v) an3 h THEMg :=true; v:=succ(v);
IFsucc(v)=rTHEnh:=true
END ELSE BEGIN
IFr=succ(v)THENr:=pred(v) delete succ (v) from CH(S); v:=pred(v)
END
UNTIL g
END.
In the algorithm, h is true iff all elements of CH(S) are examined as a middle point, while g is true iff h is true and there is an element exmined twice as a middle point. Finally r is the farthest predecessor (in CH{S)) of element to be tested next which had already a successful test (if any, or element to be tested next otherwise).
h will became true if, after one successful test, next elanent to be tested is r (i.e. the element examined first among remaining elements) . A test should be performed again on r (since the predecessor of r may change) and,if successful, will terminate the algorithm (by assigning g to true). Otherwise r is moved to its predecessor and is equal to the currently tested element until a successful test id found. The technique is simple but not trivial (L2,3, 4] failed to apply it correctly in case of GraKam's scan) .
2. The correct code of Graham's convex hull algorithm
Given à set S of n points in the plane,the convex hull CH(S) of the set is the smallest convex polygon containing the set. Graham [l] proposed an algorithm for determininrr the convex hull of a finite set of points in the plane that runs in O(nlogn) time. Variations of his method giving increasing efficiency have been described by Koplowitz and Jouppi, arid by Preparata, Lee and Shamos. In this section we give some corrections to their results. A correct code is given by means of the algorithm- for deletion of reflex elements from doubly-linked circular list.
After constructing a set of points S in polar coordinates ordered about an interior point 0 in terms of increasing angle, and initially placed in CH(S), Graham's scan algorithm [l] repeatedly examines triples of consecutive points Sj^,82,82 in counterclockwise
order to determine whether or not they define a reflex angle (an angle ^ . If an internal angle Sj^S2S2 is reflex then the test is unsuccessful and $2 cannot be extreme point because
it is internal to triangle OSj^S^. The point S2 will be eliminated from CH(S) in this case and the next test is S^S^S^. If the test is successful then $2 is a convex hull candidate and a
single scan around the ordered points will advance by checking S2S2S^.
The correct code for the Grahman's scan can be obtained by specifying■the function f (pred(v),V,succ(v)) in the algorithm in Section 1 to be angle equal to the difference between- -n and angle formed by pred (v), v and • succ (v). The angle at v is reflex iff v is reflex under our definition of f.
Koplowitz and Jouppi [2] claimed the following.
PROPOSITION. If all the points in CH(S) have been tested as a middle point and the last .test performed has been a successful forward test, then the algorithm is complete.
Tiq. ^ ■
We .show by a counterexample that this proposition is incorrect. In Fig. 1 we have the following tests :

Only the fifth test is unsuccessful. Thus only the point Sj^ will be eliminated fron
CH(S). All the remaining points S2,S2,S^,Sj have
been tested as a middle point and the last performed test S^S^S2 has been a successful forward
test. However, S2 is not a convex hull point
and hence the proposition is incorrect.
The next tests in our example are:
S5S2S3, S^SgSj, S5S3S4.
The remaining points S^jS^ and S^ form true convex hull because of successful tests
We give a correct form of above proposition.
PROPOSITION-. If all points in CH(S) have been tested as a middle point and there is a point tested twice so that all remaining points in CH(S) have been tested between these two successful testes then the algorithm is complete.
Proof. Let S^ ,82,. . .	be the remain-
ing points in CH(S) . Assume that-S^^ jSj, .. .
have been tested as middle points and the next test is successful test for This implies that there is a point S' (not necessarily on the convex hull) such that tests S'Sj^S2, S^^SjS^ ,
...,S 18„8, and 8 S,S_ have been successful. n-i n 1	n 1 ^
Thus CH(S) consists of only the extrane points
of S.
Fran the proof it follows that the proposition in [2] is correct if the original point is choosen as a convex hull point (this restriction is not mentioned in [2]).
As a consequence of their proposition Koplowitz and Jouppi have -sliöwn that 2N-M+1
38
tests are sufficient where M is the number of points in CH(S). In our example N = 5.M = 3 and we need 2M - M + 2 = 9 tests . Thus 2N - M + 1 tests will not assure completition of the algorithm unless the starting point Sj^ belongs to the convex hull.
Let us consider the exmple on Fig. 2. We have the following tests:
S1S2S3' V1S3' ^n-lVa-

There are exactly N tests in this case.
This contradict the following assertion in ,2].
"The minimal number of tests is N and occurs when no points are deleted. For each deleted point an extra test is necessary. This N - M extra tests are made making a total of 2N - M tests".	'
The correct assertion is that for each deleted point no more than one extra test is necessary. From this it follows that the number of necessary tests is between N and 2N - M +2. Hence the correct stopping rule is not as efficient as it seems in [2]. We described a more efficient stopping rule. If the points are sorted after one of the coordinates then we have a convex hull vertex for free. Thus, it is not of great interest to design a stopping criterion that depends on an arbitrary starting point because it is so trivial to start on a convex hull vertex. However Graham's algorithm sorts points in terms of increasing amie according to an interior point and finding a min or max requires additional time. Also, it is of interest in case where the convex hull of star-shaped polygon (the polygon with all vertices visible visible from an interior point) is to be obtained.
We will also discuss the modification given in [3,4]. Preparata, Lee and Shamos arrange the sorted points into a doubly-linked circular list with vertex labeled START (for instance, the rightmost smallest-ordinate point) as an initial vertex. The NEXT and PRED are associated with a node pointing respectively to its successor and predecessor in the list. The scan
Fifl, 2
terminates after advancinq all the way around to reach START. They propose the fellowing Graham's scan algorithm.
BEGIN
v:= START;
WHILE(NEXT[v] ^ START DO
IF the test(v,NEXT[v],NEXT (NEXT[v])) is successful THEN v:= NEXT[v] ELSE BEGIN delete NEXT[v] -, v:= PRED[v3 END
END.
Isolile they describe a correct rule, their code is incorrect. As an counterexample we consider the points on Fig. 1 with START pointed to Sg. The first test S^S^S^ is unsuccessful and after deleting Sj^,v points to S^ and NEXT[v] to
Sj, i.e. to START, and the algorithm stops immediately. This happens because the stopping rule (the while condition) is incorrect. We presented a correct version of the algorithm that works for any choice of the initial point STAPvT. The stopping rule reflects the proposition proved in the section.
REFERENCES
[1]	R.L.Graham, An efficient algoirithn for detemiining the convex hull of a finite planar set. Inform.Process.Lett. , 1 (1972),132-133.
[2]	j,Koplowitz and D.jouppi, A more efficient convex hull algorithm. Inform.process.Lett., 7(1978),56-57.
[3]	D.T.Lee and F.P.Preparata, Computational geometry-a survey, IEEE Trans.Comput., C-33{12) (1934) ,1072-1101.
[4J P.P.Preparata and M.I.Shamos, Computational Geometry-an Introduction (Springèr-Verlag New York Inc., 1985) .
COMPUTER PROGRAM FOR DETERMINING AN OPTIMUM SOLUTION IN LONG-TERM FOREST EXPLOITATION PROCESS
UDK 519.854/.857
Janez Barle, Janez Grad Ekonomska fakulteta Borisa Kidriča, Ljubljana
One can define the long-term forest exploitation process in different ways. Methods of operations research can be used successfully for this purpose. A group of experts at the University of Ljubljana, Yugoslavia developed a specific tree-growing function and, based on that, prescribed a set of procedures, relations and states which could represent mathematical model for optimizing the forest exploitation process. The mathod was defined as a deserete dynamic programming process, based upon Bellman's principle. In this paper we briefly describe the model and computer program prototype for solving it. They both can easily be extended by introducing more complexity into them. The program was written in FORTRAN and tested on DEC-10 computer.
1. INTRODUCTION
In the paper we describe a possible usage of dynamic programming, based upon Bellman's principle, in order to obtain the optimum solution of a predefined long-term forest exploitation process.	In this process measures and activities are introduced at each step within the iterative solution process and their effects on the intermediate results and the final solution are stated in'the form of mathematical relations. The whole forest area which is taken into consideration is divided into a number of homogeneous parts segments. For each of these segments, five special functions are imposed by means of which"we direct (quide) the process within a number of time intervals, either years or decades. The functions are: (Fl) the maximum possible tree growth capacity, (F2) exploitation capacity, (F3) the quality of the existing tree specimens, (F4) level of administration - care for improved growing conditiond, and (F5) stage in segment development, based upon the age (oldness) of the tree specimens in it. The functions help to optimize exploitation endeavours. At each step of the interative solution process four possible activities can be imposed: (Al) no activity at all, (A2) exchange of the existing tree specimen with a new one, (A3) rarefying, and (A4) restoration. The exploitation policy is defined by a sequence of chosen activities during the iterative process. The stated functions, activities and time intervals are part of the mathematical model and dynamic programming process respectively. The problem and the corresponding computer program can be extended by incorporating some additional functions and activities. Program prototype in FORTRAN was written and tested on DEC-10 computer at the University of Ljubljana, Yugoslavia. In the paper we also comment the problems which are associated with dynamic programming applied to such type of problem.
2. A PREDEFINED LONG-TERM FOREST EXPLOITATION PROCESS
We mean by that a stated sequence of policies and activities that should be pursued in forest exploitation within a life-time long period of some tree specimen in order to achieve optimum results in accordance with predefined goals and criterions. The possible and necessary actions and the most siuitable time for these actions to be carried out depend on the values of some chosen and defined functions which describe sufficiently the tree-growing process. It is these functions, which must be predefined by a group of experts in forest exploitation process - both practicians and theoricians, by means of which more or less complexity and reality is introduced in the basic model that is to be solved. After several years of analytic and experimental studies of a group of experts in Ljubljana, Yugoslavia (VADNAL, KOTAR, ZADNIK, STIRN, GAŠPERŠIČ /2/, VADNAL, ZADNIK, STIRN /3/), the following suggestions and basic ideas have been made:
(i)	The paradigm is bounded both geographicaly and in time. The whole area considered is divided into smaller regions which are further partitioned into smaller parts with specific environmental conditions and characteristics. Some parts within different regions can have similar or nearly equal environmental conditions and characteristics. They can or can not be exploited (treated) independently from each other. The area is exploited for a specific number of time periods, measured in years or decedes.
(ii)	In order to evaluate (quantify) observed characteristics of each part of the forest under consideration, time dependant analytical growth functions have been developed:
1) Function of growth
Y(t) = a(l - (1+T)e X p(-T)),
where T = (2n - 1) t"/(np"), for a > 0, p > 0, and n > 1. a defines the asymptotic value of Y(t), p = tj is the value of t
from where on Y(t)" starts to decline, and n determines the speed of convergence y(t) towards value a. The value of t = t,, when the forest restoration procedures start can be defined by solving the equation
l + T + T^-exp(T) = 0
which is obtained from the relation
t,Y'(t,) = Y(t3), or can be stated on the base of experiences.
2)	Function of increase (by growth) y(t) = Y(t)' where y(t2)' = y(p)' = 0.
3)	Function of average growth f(t) = Y(t)/t, where f(t,) = y(t,), and therefore t,Y(t,)= Yltj). J	J	3 J
4)	Function of uniform growth s(t) = mt, where s(t3) = Yttj), and s{t3)' = YCtj)'.
(iii) Five functions have been introduced which describe five possible states of forest exploitation process:
1)	The maximum possible tree growth capacity -Fl. Fl is constant during the exploitation process.
2)	Exploitation capacity - F2. F2 is a conti-nouous function, where 1 £ F2 £ 10. Its value shows the degree in forest exploitation.
3)	The quality of the existing tree specimens - F3.
This function enables us to make an assessment about the quality of the trees and whether to begin the exploitation process of some particular part of the forest or not. The parts are divided into five qualitative groups, 1 £ F3 _< 5.
4)	Level of administration - care for better growing conditions - F 4.
The value of F4, where 1 £ F4 £ 5, shows a degree of obstructing influence of unwanted tree species on growth of the wanted ones.
5)	The segment development stage - F5, based upon the age (oldness) of the tree specimens in it. Tree species are divided into four stages, accordingly:
0<F5 < t^ t3 < F5 < t^^
tj^ < F5 < tj, tj < F5 < tj, and
Tree species of different stages differ both in size and quality and can therefore be used for different purposes.
(iv) The forest exploitation process is carried out by performing a sequence of prescribed activities. The sequence order of these activities helps to optimize the process. The following four activities have been taken into consideration :
1) No activity at all. - Al.
In this case the forest develops in accor-dence with the laws of nature. The value of F2 changes only within the first two stages of segment development; it is diminished for an empirically defined quantity pr(stage, F2, Al). Similarly, the values of F3 and F4 change within the first three stages of development. They too are diminished for some empirically defined quantities pr(stage, F3,
where t^ indicates the upper time limit of the stated tree growth cycle, t^ is defined
as t^ = a/10.
Al) and pr(stage, F4,A1), respectively. Each time the value of the increment (1 for a year or 10 for a decade) is added to the previous value of F5. At the boundary points this causes a change of the stage of development.
2)	Exchange of the existing tree specimen with a new one - A2. We can pursue this activity only within the first two segment deveOopment stages. The reason for doing that is the una-dequate existing tree species. The exchange is carried out there - in those parts (segments) where the effects are most visible.
A2 exercises the following influence upon F2, ..., F5: Their values are changed only within the first two stages of segment development. F2, F3 and F4 are increased by sense empirically defined quantities pr(stage, F2, A2),pr(stage, F3, A2) and pr(stage, F4, A2), respectively, while the value of F5 is reduced to zero and the process starts from the beginning.
3)	Rerefying - A3.
The activity can only take place within the first three segment development stages. Within the first stage it is exercised by cutting down the unwanted species only what helps in quicker growth of the wanted ones. Here the activity imposes additional expenses. Within the second and third stage, in addition to the cutting down of the unwanted species, we also cut down some trees of the wanted species what helps in quicker growth of the most qualita'tive samples. Here the activity brings some profit. Due to a3 the values of F2, F3 and F4 are changed by some empirically obtained quantities pr(stage, F2, a3), pr(stage, F3, a3), and pr(stage, F4,a3), respectively. a3 has no particular impact on F5. The time increment (1 or 10) is added to the value of F5 after the time period expires .
4)	Restoration - A4.
Restoration starts at t = t3, ends at t = t4, and can take different length of time. It is characterized by wood-cutting on higher scale. After restoration is done, the forest segment under consideration passes over into the first stage. The way in which the stage of restoration is being accomplished has a great influence on the results of the forest exploitation process. The span of time t^ - t3 is generally divided into more steps. At each step the increments of F2 and F3 are computed and added to the previous values of F2 and F3, respectively. F4 does not change within A4, while the increment 1 (10) is added to F5 after each year (decade) that is passing by. F5 reduces to zero or some higher value after the restoration is over.
(v) The outcomes - R(stage, Ak). Each measure - activity undertaken at any step of the exploitation process, results and can be expressed in a form of costs and income. The resulting income of some particular step, say I(stage, Ak), for k = 1, 2, 3, 4, depends upon Fi, for i = 1, 2, ..., 5, and the size of the undertaken activity Ak. The entire costs can be divided into fixed costs FIX(stage) and variable costs V(stage, Ak). Accordingly,
R(stage, Ak) = I (stage, Ak) -FIX(stage) -
-	V(stage, Ak)
1) R(stage. Al).
In this case we have no income and no variable costs. Therefore R(stage. Al) = -
-	FlX(stage)
where FlX(stage) is some empirically obtained
quantity.
2)	R(stage, A2).
Activity A2 is carried out only within the first two stages of the exploitation process. We have some possible income only within the second stage, which can be expressed as qi(stage, A2)Y(t) where qi(2,A2) is some empirically stated weight (factor), and qi(l, A2) = 0. Fixed costs FlX(stage) are defined (empirically). Variable costs are obtained as a total of the costs of removal the unwanted species, say q^(stage, A2)Y(t), and the costs of planting new samples, say NT(stage), where q^ and NT are defined empirically. Accordingly,
RCstage, A2) = qi(stage, A2)Y(t) --FlX(stage) - q5.(stage, A2)Y(t) - NT(stage) for stage =1,2.
3)	.R(stage, A3).
Activity A3 is carried out within the first three stages. The outcome R(stage, A3) can be expressed as
R(stage, A3) = qi(stage, A3)Y(t) - FIX(sta-ge) - qc(stage, A3)Y(t) for stage = 1,2,3, where q^, FIX, and q^ are obtained empirically and qi(l, A3) = 0 . q^ and q^ have a similar meaning as in case of R(stage, A2).
4)	R(4, A4).
Activity A4 is carried out only within the fourth stage. We differentiate two possibilities:
I.	t3 = t4, when all work is done in a very short time. In this case we deal with one outcome only, defined as
R(4, A4) = qi(4, A4)Y(t3) - FIX(4) -qc(4, A4)Y(t3) - NT(4)
II.t3	^ t4, and assume that there are d4 steps of the exploitation process within the fourth step. At each step n4, where n4 = 1,2, ..., d4, we compute R(4, A4, n4) as
R(4,A4,n4) =	qi(4, A4, n4)Y(t) -
-FIX(4) -	qc(4, A4, n4)Y(t) - NT(4, n4)
where again qj^, q,;., and NT are defined empirically.
(vi) Managing the exploitation process is possible before and after the process being in progress. By this we mean some further prescribed conditions and rules which should be or should not be taken into account within the problem solving process, in accordance with the type of optimization process that we pursue. We distinguish among the following possible alternatives:
No alternations of the prescribed exploitation process are possible while the process being in progress.
Some alternations of the originally prescribed exploitation process are possible while the process being in progress. We may interrupt the process, insert some new input data and proceed the process from this point on or start the program from the beginning.
■ - The final result of the process is prescribed in advance as well as the starting conditions.
The final result is the optimum value that can be obtained by the prescribed starting conditions.
3. DYNAMIC PROGRAMMING PROCESS ALGORITHM PROTOTYPE
3.1.	General description of the deserete dynamic programming
Deserete dynamic programming process is an iterative process (BELLMAN, DREYFUS /1/). At each step i, for i = 0, 1, ..., N, we define a certain number of possible points Xj^j, for j =
1.2,	..., Mj^. For each point we compute the function value fj^j which is involved in the process of optimization. This value is the optimum value among function values fi-i k, for k = 1,
2....	Mi_i, incremented by the computed outcomes between Xj__i jj and this particular point
x^j. After computing fij for all i = 0, 1, ----
N and j = 1, 2, . . . , Mj^ we define the optimum ' sequence of the operations and decisions that were made during the exploitation process for all steps i = N, N-1, ..., 0. Sometimes two or more alternatives are possible which all give the same optimum solution.
3.2. Forest exploitation dynamic programming algorithm
In order to start the process we need the following input data:
- a, p, n, ti, t2, t3, t4; in the prototype we don't compute the values of t^, t2 and t3 but we read them as input data instead.
For each segment, i.e. for each case of the exploitation process:
all empirically defined values of FlX(stage), .NT(stage), qi (stage, Ak) , and q^(stage, Ak) , for stage =1, 2, 3, 4 and k = 1, 2, 3, 4.
all empirically defined values of pr(stage. Fi, Ak) , for stage = 1, .'. . , 4, i = 2, 3, 4 and k =1, ..., 4, that represent the increment of Fi due to the activity Ak.
F2, F3, F4, F5 which define the starting conditions of the forest exploitation process. The whole experiment can be repeated several times for different starting values of F2, ..., F5.
The process is as follows:
For each t, where t.= F5 + 10, F5 + 20, ..., in accordance, with the stage to which t belongs (stage = 1, 2, 3 or 4), the possible activities (Al, A2, A3 and/or A4) at that stage take place for all existing (active) points x^-io s (see
3.1.),	where s=l,2,3,....
For each activity that takes place at some ^t-10,s the following happens:
new values of F2, F3, F4 and F5 are computed, defining a new point	and a new step.
There is: new value of Fi = old value of Fi + pr(stage. Fi, Ak), for i = 2, 3, 4, where "-" sign appears for Al, and "+" sign appers for A2, A3 or A4, respectively. The value of F5 is increased by 10 or reduced to zero (for A2 or when F5 > t4).
the outcomes R(stage, Ak), see chapter 2.(v), arc computed and added to ft-10,s (see 3.1) in order to obtain ft,j-
Some of the 250 possible different points x^-j, for j = 1,2, ..., 250, are encountered at each step t of the iterative process. They are defined by different values of F2, F3 and F4 (10x5x5 = 250). The computer program builds two arrays X(250, 6) and FX{250) at each step of the iterative process in order to save all the the necessary intermediate data. The elements
in row j of array X contain the following data about the point xt,j:
xji = F2, Xj2 = F3, Xj3 = F4, Xj4 = k, Xjs =
= s, Xj6 = F5
where
k (=1, 2, 3 or 4) definds the activity Ak taken at step t-10 that caused a transition to xt,j, and
s definds the point xt-io,s fJ^om which the transition to xt,j made.
s stands for the row number of array X at previous step (t-10) in which the data about stored.
The elements FXj, for j =1, 2, ..., 250, represent the values of the computed ftj.
II.
After completing the first part of the algorithm we locate the optimum value fT,j i" the last step T of the interative process, where
fT,j = optimum (fT,s'	s = 1, 2, ..., 250)
Afterwards we trace back to the beginning all actions that were carried out during the exploitation process. We do this by means of data stored in arrays X and FX.
If the number of steps in the iterative process is fixed and defined, let say by t4/10, then T = t4 in case when the starting value of F5 = 0, and T < t4 otherwise.
4. CONCLUSIONS ON THE APPROACH
Obtained experiences show that a method for optimising the forest exploitation process, based upon deserete dynamic programming is reasonable and adequate. More complexity and necessary modifications can easily be introduced after obtaining and analyzing some experimental results. Large number of input data and intermediate results which are storage demanding may be regarded as the only inconvenience. Different strategies may reduce this problem by applying the secondary disk storage at each step of the interative process. The empirically obtained input data can also be stored in files in advance and kept there for as long as necessarv.
REFERENCES
1.	Bellman, R., E., Dreyfus, S., Applied Dynamic Programming, Princeton University Press, Princeton, New Yersey, 1962.
2.	Vadnal, A., Kotar, M., Zadnik Stirn, L., Gašperšič, F., "Uporaba rastnih funkcij v gozdarstvu". Zbornik gozdarstva in lesarstva 23, str. 149-179, Ljubljana 1983, Yugoslavia.
3.	Vadnal, A., Zadnik Stirn, L., Mathematical Model for Long-term Silvicultural and Utilization Planning, SYM OP IS'86, Herceg Novi, 7-10 October 1986, Proceedings, p.p. 183-189, FON, Belgrade 1986, Yugoslavia.
Fig.I - Analytical growth functions[2.3]
RAČUNALNIŠKI PROGRAM ZA DOLOČITEV OPTIMALNE REŠITVE V DOLGOROČNEM IZKORIŠČANJU GOZDOV. Proces dolgoročnega izkoriščanja gozdov moremo opredeliti na več načinov. Primerne v ta namen so tudi metode operacijskega raziskovanja. Skupina strokovnjakov Univerze Edvarda Kardelja v Ljubljani je za določitev matematičnega modela optimizacije postopka izkoriščanja gozdov razvila posebno rastno funkcijo in na temelju lete definirala zaporedje potrebnih postopkov, relacij in postulatov. Metoda je bila definirana kot postopek, ki temelji na diskretnem dinamičnem programiranju. V članku zgoščeno opišemo prototipa modela in računalniškega programa za njegovo rešitev. Model je možno razširiti z vgraditvijo nadaljnjih zahtev in pogojev. Program je napisan v programskem jeziku FORTRAN in je bil testiran na računalniku DEC-10.
SOURCE CODE ANALYSIS
INFORMATICA 4/1988
UDK 681.3:519.683
Radovan Andrejčič Univerza v Mariboru Bojan Pečeit Iskra-Delta, Ljubljana
In a software life cycle the most expensive is the phase of maintenance. A lot of costs are caused by different solutions of similar problems. The programmers source code is an exhibitional example of a variety of algorithms solving the similar tasks. In larger programmer groups there are often arranged some kind of rules for programming the source code. They are nearly always called "programming standards".
135 programs were analyzed from four programraers groups (computer centers) each consisting of 2 - 4 programmers. Statistical methods such as statistical testing, sampling, analysis of variance, etc were used as an unbiased judge. Without knowing intermediate appointments about the programming was compared the source code between programmers, programmers within groups and the code between groups. Differences between programmers within groups were surprisingly small.
Vzdrževanje je najdražja faza življenskega cikla programske opreme. Zelo veliko stroškov povzroča raznolikost reševanja podobnih problemov. V vecjih programerskih skupinah si oblikujejo neke vrste pravila programiranja, ki jih skoraj vedno imenujejo "programski standardi".
Delo opisuje analizo 135 programov iz štirih programerskih skupin (računalniških centrov) sestavljenih iz 2 - 4 programerjev, ki temelji na nepristranskih statističnih testih, vzorčenjih, analizi variance itd. Brez poznavanja dogovorov o programiranju je bila primerjana programska koda med programerji, programerji v okviru skupine in skupinami. Presenečajo nepričakovano majhne razlike med programerji v okviru skupin.
1 Introduction
Many software life cycles from different authors have been proposed. They differ in unimportant details. It is common to all of them, that the phase of maintenance is the most expensive. This phase is now the major programming activity, and very soon more programmers will be performing maintenance than development (Jones p. 35].
How to reduce costs of maintenance? Few would disagree that the quality software is not less expensive. But what is the quality software anyway?
2 Software Quality
It is as hard to define as defining a "good car driving". It is differently comprehended from a programmer to another programmer, from one manager to another etc. With the most known facets the software quality can be defined [by Arthur) as:
software quality =
F(correctness, efficiency, flexibility, integrity, maintainability, portability, reliability, reusability, testability, usabi1ity)
where each facet can be further reviewed through more criterias. For an example the maintainability can be presented as:
maintainability = F(concision, consistency, modularity, simplicity, instrumentation, seIf-documentation)
Some of these criterias are easy to measure, others are not. Everyone can explain modularity, but descriptions vary from one person to another - from equalling modularity with the structured programming, over equalling with a "no GOTO programming", to a philosophy of cohesion and coupling.
And concision and simplicity? Specter of answers is nearly unlimited. Different comprehensions cause different solutions. And this is very often a reason which makes programmers spend more time and money to understand the other programmer than to solve the problem.
Achieving an uniform coding through exact standards is not realistic. "Many rules do have legitimate exceptions" (Grauer p. 921. But on the other hand - nearly every group of programmers or computer center elaborates its own philosophy of programming. That guidelines are usually called "programming standards". So, a kind of uniformity is possible. But how much?
3 Source Code Analysis 3.1 Technics
It is of course impossible, or at least too expensive to extract data from a sample by hand. A tool or tools are needed.
Our research of the source code has based on two programs. The first one has been oriented on the analysis of the WORKING:-STORAGE section. Its input has been the cross reference and the map listing produced by the compiler. Results have given information about distributions of variable descriptions, number of references, number of words in variables, paragraphs, USAGE clauses, etc.
The other program has been oriented on a procedure division. It has produced a table of usages of the COBOL reserved verbs. Occurrences of each verb have also been analyzed in the IF statement. Logical operators have been counted detail in either IF and PERFORM UNTIL statements. This program has also given a number of comments, number of paragraphs, sections, library lines of COPY statements, total number of verbs etc.
Both programs as well as the whole research were done under the DELTA/V V2.0 operating system. Because the majority of the sample programs were written for the PDP-11 computer with the DELTA/M operating system, a little recoding was sometimes;;-, needed. What does this mean for the transportability of programs? (This interesting question is not the subject of this paper).
. 3.2 Sample 3.2.1 Criterion for a Sample
Collecting and analyzing a sample is not only a technical problem, but also an operational one. Very important question is immediately arisen: which programs to include in a sample - every program of an application or just the significant ones? In the first case, the analysis gives the exact answer about the application. But this perfection can hide differences between similar programs. It might show greater similarity than it really exists.
used.
In our research the second method has been
3.2.2 Sample Size
Four applications (programming groups) from different computer centers were included into the sample.. It was common to all of them that they used the same computer language -COBOL and each group had formed some kind of its own programming rules. It is not worth mentioning that they all sweared on the structured programming (which was prescribed in their "standards").
In this paper applications are mar)ced with letters "A" through "D" and programmers within a group with numbers. Data in table 1 have no significant meaning. They are presented just as an illustration of a sample size.
Table 1 - Illustration about the Sample Size
IProglNo p! "No of ! Average IStahdar! Exec. ! Iramrlrgmsl. lines ! 1in/progr!deviat.! verba !
! Al	]	11	!	5274	!	479.45
! A2	!	10	!	5517	!	551.70
A3	!	5	!	1344	!	268.80
A4	!	9	!	4550	!	505.56
Bl	!	16	!	13977	!	873.56
B2	!	17	!	17706	!	1041.53
83	!	2	!	2548	!	1274
CI	1	15	1	10599	!	706.60
C2	!	10	!	6437	!	643.70
Dl	!	22	!	45908	!	2086.73
D2	!	8	!	15606	!	1950.75
D3	!	10	!	11149	!	1114.90
sA	!	35	!	16685	!	476.71
! SB	!	35	!	34231	!	978.03
! sC	!	25	!	17036	!	681.44
! sD	;	40	!	72663	!	1816.58
ISUM	!135	1140615	1	1041.59
277.0 331.6 180.0 209.4 349.6
555.2 393 195.9
323.1 617.6 775 . 2
389.3 281.9 475.8
256.4 ! 731.2
! 731. "9
!	2544	!
!	3079	!
!	706	!
!	2908	!
!	5116	i
!	6984	J
!	1029	!
1	3042	!
!	2187	!
!	19147	!
!	6834	!
1	3518	!
!	9237	!
.1	13229	!
!	5229	!
!	29499	!
!	57194	!
3.3 Analyzing Comment Statements
Comment
3.3.1 Importance of the Statements
"Although COBOL is often thought of as a self-documenting language, this is only partially true. With a careful choice of words,	each statement can indeed be
self-documenting, but it cannot explain its own purpose: it merely states its contribution to a technique or algorithm" [Ledin, Kudlik, Ledin p. 97].
Comments are still needed, they become even more and more important. Specially in the last time, when programs are often not maintained by the original author. As Yurdon says "No programmer, no matter how wise, how experienced, how hard pressed for time, no matter how well intentioned, should be forgiven an uncommented program".
3.3.2 Number of Comments per Source Code
c
Absolute number of comments in a program does not have any meaning. It needs to be compared with the number of source lines, or the number of executable statements, or with the reserved COBOL verbs. Table 2 presents data about the number of source lines per comment where source lines per comment (SLC) is calculated as
SLC =
total number of lines
number of comments
Table 2 - Source Lines per Comment
IProglNo p! No of ! Average 1ramr!rgms!comment 1 SLC
IStandar! Sum of ! ideviat.! squares !
! Al !	11 !	257 !	20.54	5.76	! 5003.5 !
! A2 i	10 i	318 !	17.35	3.48	! 3131.2 !
! A3 !	5 !	95 !	14.14	4.43	! 1097.7 !
! A4	9 !	223 1	20.37	5.90	! 4046.0 1
! BI	16 !	1820 1	7.68	2.52	! 1044.3 !
! B2	17 !	2720 !	6.51	1.54	! 761.1 !
!• B3	2 !	432 1	5.9	1.50	! 74.1 !
! Cl	15 !	2611 !	4 .-06	0.49	! 250.8 !
! C2	10 !	1387 !	4.64	1.63	! 241.7 !
! Dl ■	22 !	8140 !	5.64	0.42	1 702.8 !
1 D2	8 !	2922 1	5.34	1.22	1 239.9 !
! D3	. 10 !	2418 !	4.61	0.5	! 215.1 !
	:======	=======	==========	::= = = = = = =	: = = = = = = = = =:=: +
3.3.3 Differences in Percentage of Comments between Applications
Is there any significant stability in commenting programs?	This answer was
researched with the analysis of variance. In table 3 there is an analysis of not only differences between programs and programmers, but also of differences between applications. [Andrejcic p. 1611,
Table 3 - Analysis of Variance between Applications and Programmers
1 Source of IDeglSum of! Mean ICalcu- ! F ! ! Variation ! fre ! square ! square !lated F! tablel
lApplications! 3 ! 4506.4 ! 1502. 1 ! 37 .33 ! 23.70! !Programraers ! 8 ! 207.3! 25.91! 2.92 ! 2.663! ! Programs !123!1091.2! 8.87!	!	!
The first null hypothesis - that differences between groups (applications) are not significant can be absolutely rejected '^(3-0.OQl) " 23.7).	The second null
hypo£hésis, that	differences	between
programmers do not exist can be rejected too, but the risk is this time for a bit greater over half a per cent '^(a-O.Ol) ~ 2.663). This has given a reason for a "àetati ed investigation about differences between programmers within groups. (See table 4).
Table 4 - Analysis of Variance between
Programmers within Applications +=========================================+
ISource !App!Deg! sum !mean!calcula! F(0.1)! !of var.!lic!fre!squar!squ.!ted F ! table!
Ibetween!	A !	3 !184.3!61.4!	2.125 !	2.28 !
1 prog !	B !	2	! 13.7! 6.9!	1.502 !	2.49 !
I ramm !	C !	1	! 2.0! 2.0!	1.549 !	2.88 !
I ers !	D !	2	! 3.62! 3.6!	7.262 !	2.44 !
Ibetween!	A !	31 ! 896.2 ! 28.9 !	!
! prog !	B !	32!146.6! 4.6!	!
! ramms !	C !	23! 30.0! 1.3!	!
!	!	D !	37! 18.5! 0.5!	!
f ^- = = =------■ ■ : ■ _ :i	_ _ _	_ .	; : - ■ - r_ - zz - t
These analyses have shown, that the only application in which significant differences exist was the application "D" (for the risk of 10%, but it is not greater for the risk of 0.1% - F(2-0.001) " 7.29): The t-test proved that the programmer "D3" had more comments than the other two. Results have shown greater stability in writing comments than it had been expected.
Comparing the average (9.03 source lines per comment) of this sample with the previous investigation gives also unexpected results. Al-Jarrah-Torsun (page 344) have • counted an average of 66.6 source cards per comment card. It is such a great difference, that it needs no special statistical prof. It does not also need the result of Smolej-Korelic - 23.82 lines per comment.
3.3.4 Correlation between Program Length and Number of Comments
Naturally, it is expected that longer programs are more complex and for this reason they need to be more commented. But the previous investigation of dr. Smolej and Korelic(*) discovered unexpected negative correlation	.between	comments	and
characteristics of complex programs.
Table 5 - Correlation between Program Length and Density of Comments
+ = = = = =j=: = = = = = = = = = = = = = =:5= = = = = = = = = = = = = = s: = = = =: = = = +
! Prog!Coe.cor.! t !	t(table)	!
!	Al	!	0.160	! 0.486	1 t(0.50; 9)	=	0.703	!
t	A2	!	-0.186	! 0.525	! t(0.50; 8)	=	0.706	1
t	A3	!	-0.702	! 1.707	! t(0.10; 3)	=	2.353	!
!	A4	!	0.771	! 3.199	! t(0.01; 7)	=	3.499	!
!	BI	!	-0.186	! 0.708	! t(0.40;14)	=	0.868	1
!	B2	t	0.148	! 0.580	! t(0.50;15)	=	0.691	!
!	B3	t	1.000		I			!
)	Cl	!	0.277	! 1.038	! t(0,20;13)	=	1.350	1
!	C2	1	-0.535	! 1.790	! t(0.10; 8)	=	1.860	!
f	Dl	t	0.304	! 1.428	! t(0.10;20)	=	1.725	1
!	D2	!	0.705	! 2.435	! t(0.05; 6)	=	2.447	1
1	D3	!	0.409	! 1.269	! t(0.20; 8)	=	1.397	!
+ :		: u =	========		:============:	= = :	= z = = = = =	: +
According to table 5 programmer "A4" is the only one who can be assumed to have larger programs less commented. The risk of rejecting the null hypothesis, that the correlation coefficient is not significant, is about over 1%. The nearest result of the programmer "D2" increases this risk up to over 5%.
Four programmers ("A2", "A3", "Bl" and "C2") had even negative coefficient correlation. This means that the larger programs had relatively more comments. Programmer "A3" had this coefficient even -0.702, but his amount of the sample (5) was too small to reject the null hypothesis. Programmer "C2" with the coefficient correlation -0.535 and greater amount of sample (10) was much closer to the rejection of the null hypothesis.
3.3.5 Sampling Contents Of Comments
"The mere presence of comments, however, does not ensure a we11-documented program, and poor comments are sometimes worse than no comments at all" [Grauer p. 103].
There are also known the first rules which suggest how to write comments (to explain reason, not to duplicate code, etc). Their present usage can be compared with the first considerations about structured programming in the early '70.
How to establish the quality of comments? At least two problems occur. The first one is to distinguish good comments from bad ones. It is impossible to do it automatically. A man as an observer and arbiter is needed. And this causes the second problem. The amount of comments is too large to examine every comment line.
Dr. Smolej and Korelic have analyzed 238 programs written by 8 programmers from one computer center with the goal to find representative characteristics of an average program.
Sampling was chosen to give an illustration of the quality of comments. Samples of comments were collected from each programmer. Each line was subjectively estimated as good or bad. Criterias for a good comment line were easy to satisfy. Each line that might be of any help in understanding the code was accepted as good. No additional comparing with the nearest lines of source code was done. Also the AQL was very low - 10% with the risk of 5%. MIL. STO. 105D double sampling plan was used.
Table 6 - Sampling the Quality of Comments
!Prog!
Irarar! +----+
1	Al	!
J	A2	!
J	A3 !
1	A4	1
!	B1	!
1	B2	1
!	B3 !
!	CI	!
!	C2 !
1	D1	!
I	D2 I
!	D2	!
Numb.! First comm.! amo.AC
sampleISecond RE reslamo.AC
257 1 318 !
95 ! 223 ! 1921 ! 2720 !
432 i 2610 ! 1378 ! 8140 ! 2923 ! 2418 !
20 32 13 20
80	11
80	11 32 5
80	11 80 11
80	11 80 11
80	11
7 4	1
9 14	!
5 4	i
7 12	!
16 46	!
16 39	!
9 17	!
5	!
20 8
13 6 7 10
sample !Acc RE resiRej
9 7 !	AC
!	RE
!	RE
!	RE
1	RE
!	RE
!	RE
!	AC
!	AC
!	AC
!	AC
!	AC
Results have very clearly rejected programmers with the less commented programs and accepted prograiramers with better results. What coincidence? Obvious ly, some groups take a great care about this problem, while the others do not!
3.4 Analyzing User-Defined Words
In applications "C" and "D" similarities are immediately seen. Not only tests of the mean values and F-tests of the intermediate differences of standard deviation, but also much harder the Kolmogorov-Smirnov test of goodness of fit have proved that there were no differences in	distributions	between
programmers within groups. This was specially surprising in the application "D", where distribution of each programmer was bimodal. (Every programmer had more variables with length of 13, 14 or 15 characters than with 10, 11 or 12). The first explanation was, that this was caused by the influence of the "COPY" statements. But further analyses had contradicted this suspicion.
348]
The Kruskal-Wallis procedure [Andrejcic p.
12
H =
1369
12
100
<12 + 1)
361
144
•> -
- 3
3	2	3
(12 + 1) = 7.47
with chi-square
7.81
M0.05;3)
no reason to reject the null hypothesis.
compared gave
that there were no significant differences between applications. However, the result was very near to the border value for the risk of 5%.
4 Conclusion 4.1 Interpretation
Beside correct comments a mnemonical significant data names are very important for understanding a data flow. Nearly all authors who deal with programming techniques suggest to use as many of the 30 characters as needed to make names in a program easy to understand. Not only to the original author, but for others as well.
Maybe this is a reason for a surprise when Al-Jarrah-Torsun discovered that the average iiser defined name had "only" 7.81 characters. In the next table distributions of all user-defined names are shown. Results are grouped into classes of three lengths. The hyphen is counted as the other characters.
Only programmers "Bl" and "B2" have user-defined names longer than 18 characters.
The first discovery was that the relative number of comments is increasing (comparing with the oldest analysis by Al-Jarrah-Torsun and a bit younger by Smolej-Korelic). All applications were produced with the interactive editor, while Al-Jarrah-Torsun wrote about cards. So, maybe also the economical effects can have some influence on the density of comments.
Not only the density, but also the constancy was surprising. It was even not effected by the program length, as it had been measured by the previous analysis. Influence of the group agreements on the programmer were reflected immediately. This brings to a conclusion, that commenting is given more and more care. It has now its place also in "the programming standards".
Table 7 - Distributions of Lengths of the User-Defined Names
! Length !	Al ! A2 !		A3	1	A4	! BI	1 B2	I	B3	[	C1	1 C2	I D1	1 D2	1 D3	
! 1-3 !	100 • 66 !		95	1	248	1 96	1 19	I	15	1	384	1 272	I 180	1 59	1 30	1
! 4-6 !	334	! 548 !	105	!	333	! 596	1 995	1	308	i	1643	11026	12168	I 813	I 731	1
1 7-9 !	419	! 270 !	82	!	323	11070	11400	1	266	I	341	1 221	13966	11269	11072	I
! 10-12 !	72	! 153 ;	4	t		1 315	1 760	1	58	I	76	•1 76	1 385	I 238	I 179	1
! 13-15 !	22	! 50 !		t		! 255	1 301	I	9	I	30	1 10	11155	1 376	1 296	t
! 16-18 !		11 i		!		! 186	I 126	i	2	t	2	1	1 21	1 16	1 13	I
! 19-21 !		1		!		1 142	1 38	1		1		I	1	1	I	1
! 22-24 !		t		1		! 105	1 . 1	1		1		!	1	t	1	t
! 25-27 !		!		1		1 94	1	I		!		I	1	1	1	!
! 28-30 !		t		1		! 44	i	1		1		1	1	I	1	1
! average !	6.6	7.2 !	4.9	- + -1	5.2	110.7	I 8.7	■ + -I	6.9	- + 1	5.6	I 5.5	I 7.9	1 8.0	1 8.2	- + i
1 rank !	8	6 !	12	1	11	! 1	I 2	1	7	1	9	1 10	I 5 :======	! 4	1 3	I
Sampling of comments gave some disappointing results, or at least unexpected. It was very easy to distinguish between the good and the bad comments. Criterias were easy to achieve, but results rejected programmers with the less commented programs.
After the research was finished, each "programming standard" was studied in detail. Results of the analysis were compared with these prescriptions. In applications"C" and "D" detailed programming guidelines about the form of comment were stated, while in others they were omitted.
Al-Jarrah-Torsun found that the average user-defined name had 7.81 characters and their expectation that it "was expected to find them to be on average much longer" [Al-Jarrah-Torsun p. 343] was not in place. It seems that an average of 8 characters is the most common value. Equality of distributions of the user-defined names was greater than expected. Descriptional estimates about the importance of the long user-names were:
Bl - very important B2 - very important B3 - very important CI - less important C2 - less important Dl - very important D2 - important D3 - important
An interview with programmers on the application "A" was not possible. Answers were as expected, except the programmer B3's and Dl's. Programmer "83" was a beginner and the worst typist. "D3" was also very bad, the worst in his group, but they both answered under impression of the group agreements. If the programmer "B3" would be separated, the Kruskal-Wallis procedure
12
h =
11 » (11 + 1)
1156
289
144
■) -
- 3
(11 + 1) = 8.If
would reject the null hypothesis (with the risk of 5%), that there were no differences between applications about lengths of the user names. This would prove, that the statistical significant differences exist. For this reason the correlation between the typing speed and the length of user-names was not analyzed. As there were nearly no differences between programmers within groups, results were obviously more depended on agreements than the dexterity. The suspicion, that the uniformity of distribution was caused by the COPY statements in the WORKING-STORAGE section was comprehended. The amount of user-names from the library files was found to be very low.
With the method of comparing the mean value with the constant it was evidenced that each programmer had different average than the sample of Al-Jarrah-Torsun (7.81) with the greatest risk of 3.67 for the programmer "D2". Coincidently, the whole sample together had an average of 7.79 with standard deviation of 3.62, so critical risk (CR) lAndrejcic p. 1001
7.81 - 7.79
CR = - = 0.916
3.62
gave no reason to contradict the hypothesis that the both samples had statistically equal mean values.
4.2 Comment of the Analysis
It needs goal of this quality of the similari t ies appiicationa application. short illustra part of the course more ca
to be stated clearly, that the analysis was not to point to the software. The goal was to find and	differences between
and programs within an And this paper is only to give a tion of the analysis, so only a research is shown. There are of Iculations and comparisons.
This analysis neither measures nor estimates the quality of applications. It is impossible to do it just on some facets about the state of the source code. It is well known, that the quality of the software is designed and determinated in the previous phases of the software life cycle.
The quality of the source code is not the most important component of the software quality. So, it cannot be made equally with the software quality which make part of the linear equation [by ROLAND]
x = wi
+
+ C
\f 27464
where W's are weighting factors and X's are software metrics - each of which may be or may not be given in turn by linear equations of the same form, and C is the constant. One of them is also the maintainability as it had been shown at the beginning. Uniformity of code can be of a great help in eliminating difficulties and frustrations in authorship of the program.
Anyhow, the analysis proved that differences within groups were very small, but "programming standards" were different. Even if they all referred to the same philosophy, they are a great reason for different solutions of similar problems. And our opinion is that this is a subjective argument, which needs to be eliminated. Our analysis examines for the realistic possibilities to achieve it.
References
dr. Radovan Andrejcic: "STATISTIKA PRI KADROVANJU IN IZOBRAŽEVANJU" - VSOD Kranj 1979
Lowell Jay Arthur: "MEASURING PROGRAMMER PRODUCTIVITY AND SOFTWARE QUALITY" - John Willy & sons 1985
dr. Robert T. Grauer: "STRUCTURED METHODS THROUGH COBOL" - Prentice Hall Capers Jones: "PROGRAMMING PRODUCTIVITY" -McGraw-Hill 1986
M. M. Al-Jarrah and I. S. Torsun: "EMPIRICAL ANALYSIS OF COBOL PROGRAMS" -Software Practice and Experience 9/1979 George Ledin Jr., Michael Kudlick, Victor Ledin: "THE COBOL PROGRAMMER'S BOOK OF RULES" - Belmont California
Dr. Vitomir Smolej and Igor Korelic: "EMPIRIČNA ANALIZA COBOLSKIH PROGRAMOV" -Informatica, Ljubljana oct. 1981 John Roland: "SOFTWARE METRICS" - Computer Lang. 6/1986
dr. Francis J. Wall: "STATISTICAL DATA ANALYSIS HANDBOOK" McGraw-Hill 1986
PREVODJENJE PRAVILA GRAMATIKE IZ EBNF ZAPISA UBNFZAPIS
UDK 519.682.1
Mirjana Ivanović, Zoran Budimac Institut za matematiku
TRANSLATION OF GRAMMAR RULE EBNF DESCRIPTIONS TO BNF ONES: The use of
comput-ers in grammar processing of any kind demands the simplest, possible description of grammar rules. Backus Normal Form <BNF) is acceptable solution.
However, Extended Beckus Normal Form CEBNF) is more suitable for humans. For purposes of efficient computer processing of grammar rules described by EBNF, preprocessor becomes a must. It translates EBNF description of grammar rules to equivalent BNF description,
SAŽETAK: Obrada gramatika Cprogramskih Jezika) na raćunaru zahteva Sto jednostavniji n.aein zapisivanja njihovih pravila. Backusova normalna forma <BNF> Je Jedno od prihvatljivih reSenJa.
Medjutim, za ćoveka koji zapisuje pravila pogodnija Je proširena Backusova normalna forma CEBNF>.
Da bi se pomirila ova dva protivreCna zahteva realizovan Je predprocesor koji obavlja prevodjenje pravila iz EBNF u BNF zapis.
1. Uvod
Većina programskih Jezika se može, skoro u potpunosti, definisati pravilima kontekstno slobodnih gramatika, koja su oblika:
a ■.:= ß
metaproraenlJ i va, a desna strana Je metaizraz -niz metapromenlJ i vi h i/ili metakonstanti razdvojenih univerzalnim inetasimbolom |.
Metasimbol | povezuje različite desne strane pravila sa istim levim strajiama:
Leva strana pravila Je Jedaii od simbola iz skupa neterminalnih simbola gramatike, a desna strana pravila Je konaCan niz terminalnih i/ili neterminalnih simbola.
Potreba preciznog formalnog definisanja prograunskih Jezika uslovila Je pojavu različitih notacija	za zapis pravila
gramatike, od kojih se najviSe koriste:
a)	Bekusova CBNF> i proSirc-na Bekusova
<EBNF) notacija,
b)	obrnuta i modifi kovana obrnuta notaci ja,
c)	grafička notacija,
d^ sintaksni dijagrami.
U slučaju a> i b) sintaksa Jezika se definiSe pomoću konačnog skupa metaformula.
Metaformula se sastoji od leve i desne strane razdvojene univerzalnim metaslmbolom :: = .	Leva	strana	inetaformule	Je
a : := ßi a : : = ftz
a : : = ßr,
<=> a : ßi \ ßz \ ... \ ßn
Z. Primena BNF i EBNF u zapisivanju pravila granvatike
Da bi se omogućila JednoznaCnost pri definisanju simbola gramatike uvode se ograniCavači za obe kategorije simbola. Neterminalni .simboli se ograničavaju znakom < Cna početku) 1 > Cna kraju), a terminalni znakom	Ovakav način obele±avanJa stvara
dodatiiu opreznost i napor pri zapisivanju pravila, ali daje veće slobode kod imenovanja simbola.
Za zapisivaiije pravila kontekstno slobodnih gramatika, BNF nudi sledeću pogodnost: ukoliko postoji vlSe pravila sa različitim desnim, a
50
Jednakim levin st.i-anama, tada se ona "zapisuju kao Jedno pravilo čija Je leva st. rana Jednaka levoj st-raini niza pravila, a desna strana sadrži sve desne strane niza pravila razdvojene mstasimbolom | Cu znaCenJu ill>.
Ovakav naCin zapisivanja Je nefleksibilan i dovodi do dupliranja i nepotrebnog ponavljanja pojedinih nizova simbola u pravilima.
EBNF donosi znaCaJna proširenja Jer pored metas imbola koje koristi BNF C::"= i |> uvodi parove netasimbola:	[), <>.
2.1 Metaslmboli C> <Grupisanje)
Desne strane pravila sa istim levim stremama se nekad razlikuju samo po nekom nizu terminalnih i/lli neterrainalnih simbola. Pomoću metasimbola <> uvodi se grupisainje sa sledečim znaAenJ era :
BNF
P : : «	il ri	i2
P : : -	il TZ	iz
P	il rr, Ì2
EBNF

il < n I ■ • • I rri > I2
Bar Jedan niz simbola ii ili iz nije prazan. Takodje, ri, ... ,rn Cza n>l) nisu prazni nizovi simbola.
2.2 Metasimboll I ] CJednostruka pojavai
Dva pravila gramatike se mogu razlikovati samo po pojavi nekog niza simbola u Jednom od njih. Da bi se izbeglo pisanje oba pravila uvode se metasimboli [3 sa sledećim znaCenJem:
BNF
P : : = ni n nz P : : = m nz
<=>
EBNF
P : : = m I nI nz
Niz terminalnih i/i li neterminalnih simbola ograniCen metasimbolima t J javlja se u pravilu najvise Jedanput.
Da bi pravilo imalo smisla niz simbola n nije prazain.
BNF P P P
= ni na = ni nz na = m nz nz na
EBNF
<=> P : : «■ m < nz > ns
P : : = ni nz ... nz na
Da bi pravilo imalo smisla niz simbola nz nije prazan.
Moguće su sledeće pojave parova metasimbola
r ] i <>;
1> !...<...>..,] 2> <. . . r. . . ). . .>
3> <...<...>...> 4j	]... ]
do proizvoljnog nivoa dubine. Pojava metasimbola O je moguća na svim nivoima.
3. Prsvodjenjo pravila Iz EBNP zapisa u BNF zapis
Razvijanje prevodilaca za vlSe programske jezike Je jedna od znaCaJnlh oblasti računarskih nauka, stoga je potrebno na pogodan naCin izraziti sintaksna pravila jezika koriSćenJem gramatika.
Za obradu gramatika na raCunaru pogodniji je zapis pravila u BNF notaciji. Sa stanovišta Coveka, jednostavnije 1 prirodnije je pravila zapisivati u EBNF notaciji.
Pri rućnom prevodjenju pravila uoCavaJu se odredjene zakonitosti koje je moguće lako programski simulirati.
3.1. Zamena metasimbola <> Cgrupisanja>
Deo pravila ograničen metćisimbol ima Ci), se može zameni ti uvodjenjem novog netermlnalnog simbola. Za uvedeni neterminalni simbol Ckoj 1 Je različit od svih neterminalnih simbola gramatike) definiSe se novo pravilo:
2.3 Metasimboli C> C višestruka pojava)
Pravila, u kojima se neki niz simbola gramatike javlja proizvoljan broj puta, se u BNF ne mogu direktno predstaviti već se koristi rekurzija. Rekurzija se otklanja uvodjenjem metasimbola <> sa sledećim značenjem:
P : : = dl C pi I pz I . . . | pn ) dz
<=>
P : := dl <zamena> dz <zamena) : : => pi | pz | ... | pn.
gde su P 1 Czamena) neterminalni simboli gramatike, di i dz su nizovi simbola od kojih Je bar jedan neprazan.
3.2. Zamena metasimbola I ] Cjednosti-uke pojave)
4. Realizacija
Od poCet-nog pravila se formiraju dva nova:
- potetno pravilo sa jednom pojavoin niza term inai ni h iXl 1 i net-er-mi nal ni h simbola ograničenih metasimbolima li).
Navedeni algorit-mi za zamenu metasimbola EBNF su osnova na kojoj se zasniva rad predprocesora. Da bi se oni uspeSno i efikasno implementirali izabrana Je pogodna forma za eksternu i internu reprezentaciju pravila.
- potetno pravilo bez niza terminalnih i/ili neterrainalnih simbola ograničenih nietasimbol ima [ i } .
P : : = pi r n ] p2 < => P : : = pi pz
P ; : = pi n pz
4.1 Eksterna reprezentacija pravila
Pravila gramatike se zapisuju u datoteku. Potrebno je prilikom njihovog zapisivanja poštovati odredjene zahteve da bi se Sto jednostavnije koristila u obradi. Jednoznačno se identifikuju tri kategorije simbola:
3.3. Zamena metasimbola {> CviSestruke pojave>
Zamena niza terminalnih i/ili neterminalni h simbola ograničenih metasinibol i ma <> se mo2e realizovati na viSe načina. Međutim, u večini slučajeva zamena je uglavnom uslovljena tipom pravila i ne predstavlja univerzalno reSenJe.
Analizirajući nedostatke takvih zamena i samu definiciju višestruke pojave, dolazi se do opSte' metode za zamenu višestruke pojave koja se mo±e Jednostavno realizovati.
Od izvornog pravila formiraju se četiri nova;
Neterminalni simboli su ograničeni
znakom < na početku i znakom > na kraju. Naziv simbola se mo±e sastojati od viSe reči. Ukoliko
su reči razdvojene sa v iSe praznih mesta uzima
s& u obzir samo Jedno. Praznine izmedju < i
prvog znaka simbola i poslednjeg znaka simbola
i > se ne uključuju u simbol tj . nisu od
značaja.
- Terminalni simboli su ograničeni znakom i na početku i na kraju. Niz znakova koji predstavlja simbol se koristi u istom obliku kako Je i zapisan.
početno pravilo bez niza ograničenog raetasimbolima < i >,
simbola
u početnom pravilu se niz simbola ograničen metas i mbo 1 i ma < i >, zamer»juje novim neterminaini m simbolom.
- novouvedeni neterminalni definiSe pomoču dva pravila
s i mbo1

P : : = m < nz > na
P : : = ni nz P : := m <zamena> nz . <zamena>: :=<niz zamenjenih simbola> |
<niz zamenjenih s imbola><zamena>
Može se učiniti na prvi pogled da šu neka pravila suviSna i da bi se mogla uvesti Jednostavnija zamena. Medjutim, Jednostavnija zamena bi dovela do pojave praznog pravila Ctj. pravila sa levom, ali bez desne strane> koje može da oteža proces rada prevodioca.
-	Metasimboll se navode bez ograničavača.
Pravila moraju da zadovolje sledeče zahteve:
-	maksimalna dužina simbola zajedno sa ograničavačima Je 40 znakova,
-	jedno pravilo se može nastavljati.u viSe linija,
-	iza metasimbola := mora da sledi bar jedan simbol desne strame pravila,
-	Jedan simbol .se ne može nastavljati u viSe linija,
metasimbol | Cukoliko je potreban? stavlja se na kraj tekuće linije, a ne na početak sledeče,
-	izmedju susednih simbola metajezika i susednih terminalnih simbola objekt Jezika obavezna Je upotreba bar Jednog praznog mesta.
Tabela simbola
Naziv
Naziv
Nazi v
Slika 4.1. OpSt-i izgled tabele simbola
dok u svim ost-alim sluòajevima nije,
■netasimbol i <,), I, J, <,> se mogu ugnezdlt-i do proizvoljne dub i ne >
- za svaki met-asimbol C, t , < u tekućoj liniji mora da postoji njegov odgovar-a J uć i netčisiiiibol J, >.
U procesu uCitavanJa pravila i formiranja njihove interne reprezentaciJe vr&i se odredjena logiCka kontrola ispravnosti zapisa i ukoliko se otkriju neke greSke (nepoštovanjem postavljenih zahteva> predprocesor zaustavlja rad i izlistava pogreSno pravilo.
4.2. Interna reprezentacija pravila
Za smeStanJe pravila u memoriju nćikon njihovog ućitavanja, koristi se tabela simbola. Tabela simbola Je niz od m pokazivača diixainiCke strukture - liste koje sadrže slogove tipa
naziv. Svaki simbol gramatike se u tabeli _ simbola javlja samo Jednom <slika 4-l>.
tabela_simbolai=arrayI 1. . n] of naziv;
nazi v=record
ime_simbola:stringi40J; strukturo_pravi la: 'Opravilo; sledeci_nazi v: "^nazi v ;
end;
Slog naziv se sastoji od polja:
- iiné_simbola Je niska du±ine 40 znakova, u koju se upisuju nazivi simbola gramatike zajedno sa ogranifiavaeima,
-	struktura_pravila Je pokazivač na slog tipa pravilo, kojim se predstavljaju pravila za netorminalne simbole, dok Je za terminalne simbole vrednost pokazivača nil,
-	sledeci_naziv Je pokazivač na sledeCi slog tipa naziv u listi.
tabela simbola
nazl V
Slika 4.2. Struktvjra sloga naziv
r aizdva Jac
i
pr avi 1 o
Slika 4.3. Strukt-ura sloga pravilo
Za svaki neterminaini simbol iz tabele simbola vezuje se slog t-ipa pravilo kojim se opisuju sva pravila sa istom levom stranom.
pravi losarecord
1evi_deo_p : "naziV ;
razl icite_desne_strane: "^razdvajao; sledece_pravilo:"pravi lo ;
end ;
Slog pravilo se sastoji od polja;
levi_dep_p je pokazivač na naziv neterminaJnog simbola Cu tabeli siiiibola> koji se nalazi na levoj strani pravila,
- razlicite_desne_strane Je pokazivat na slo^ tipa razdvajac pomoću kog se opisuju sva pravila sa istom levom stranom levi_deo_p.
-	sledece_pravHo Je pokazivač na sledeüu grupu pravila.
Slog	tipa	razdvajac	omogućuje
predstavljanje svih pravila koja pripadaju istoj grupi tJ. imaju istu levu stranu,
raždvaJac=record
desne_strane:"desni_elementi ; siedeci_razdvajac:"razdvajac;
end ;
Slog tipa razdvajac se sastoji od polja:
desne_strane Je pokazivat na listu simbola desne strane pravila predstavljenih listom elemenata tipa desni_elementi,
-	sledeci_razdvajac je pokazivat na sledeći slog tipa razdvajac za koji se vezuje desna strana novog pravila iste grupe.
Slika 4.4. Struktura sloga razdvajac
Slog tipa desni_elemenli omogućuje predstavljanje pojedinačnih simbola desnih strana prav ila.
desni _element1=record
ime_elementa:"naziv; sledeci_element:"desni_elementi ; end ;
Slog desni_eleinenti se sastoji od polja:
-	iine_elementa je pokazivat na naziv simbola upisanog u tabelu simbola,
-	sledeci_eieinenl Je pokazivat na sledećl simbol desne strane pravila.
Slika 4.S. St-ruktura grupe pravila
4.3. Programst-a implementacija
Predprocesor je realizovan u Turbo Pascal jeziku kao niz procedura i funkcija, Celokupna aktivnost, predprocesora se moSte podeliti u nekoliko celina:
iz ulazne datoteke učitavaju se redom linije, odvajaju se pojedinačni terminalni i neterminalni simboli, upisuju se u tabelu simbola i formira se interna struktura pi^avila.
EBNF i BNF notacije su pogodan mehanizam za zapisivanje pravila i Široko se koriste u svim oblastima gđe se zahteva formalizovan natin zapisivanja pravila.
Realizovani predprocesor kao gotov, celovit softverski proizvod realizovan je na Turbo Pascal jeziku na IBM PC kompatibilnom raCunaru. Može se jednostavno uključiti u Sire softverske proizvode kao gotovo orudje za lakSu obradu i upotrebu pravila proizvoljne kontekstno slobodne gramatike.
- aJco se u analiziranoj liniji nalazi niz simbola ograničen metasimboli ma C,> taj niz se u pravilu zamenjuje novim neterminalnim simbolom, a izdvojeni niz se upisuje u listu zćunena kao novo pravilo.
Literatura
1. Ories D. - Compiler construction for digital computers, John Wiley & sons INC., New York, 1971.
nakon - u4i tavanja svih pravila iz datoteke analiziraju se izdvojena pravila u listi zamena,
- nakon toga koriSeenjem algoritama zamene metóisimbola I,I, <,> vrSi se njihova zamena uz logieku kontrolu (čuvaju se sve zatnenjene niske simbola da se ne bi ista niska simbola zameni la vise puta različitim neterrai nal nira si raboi i n)a>.
Ponavljanjem ovog postupka dobija se interna reprezentacija pravila u BNP notaciji.
2.	Ivanovie Mirjana - Programska implementacija sintaksnog anaizatora programskih Jezika zasnovanog na metodi prioriteta. Magistarski rad. Novi Sad, 1908.
3.	Jensen K. i Wirth N. - Pascal user manual and report, Springer-Ver1ag, New York, 198S.
4.	Tremblay J.P., Sorenson P.Ö. - Compiler writing, McGraw-Hill Book Company, New York, 198S.
6. Zaključak
U oblasti konstruisanja prevodilaca za viSe programske jezike značajnu ulogu igra izbor pogodnog načina zapisa pravila gramatike jezika.
QUO VADIŠ INFORMATIZACIJA POSLOVNEGA ODLOČANJA
INFORMATICA 4/1988
UDK 519.688
SamoBobek
Visoka ekonomsko-komercialna šola, Maribor
Priiujoći prispevek Dbravnava raiunalnièko podprto poslovno odločanje kot sinergijsko Cso)de-' 1 ovanje ćloveka in raiunalnika,ki ju v tem primeru povezujejo i n-f ar mac i jsk i oziroma odločitveni procesi, ki se odvijajo v poslovnem sistemu. V tej vlogi naj računalnik predvsem uspeènd in učinkovito podpira človekove spoznavne (kognitivne) procese in ravno ta.ka tudi komuniciranje med posamezniki.
QUO VADIŠ INFORMATIZATION IN BUSINESS DECISION MAKING; This paper describes computer supported (business) decision making as : cooperati on (with sinergy a-f-fect) betwen humans and computers, which are in interaction through decision making process in organization. Computer should-support, as part o-f such system, human cognitive process and communication between individuals in decision groups.
1 KAKŠNE RAČUNALNIŠKO PODPRTE POSLOVNE INFORMACIJSKE S.ISTEME POTREBUJEMO ?
, .	Računalniško podprta poslovne infor-
macijske sisteme predstavljamo v smislu njihovih posebnosti in značilnosti najpogosteje skozi: (1) avtomatsko obdelavo podatkov (AOP.) —>
(2)	upravljalni i n-f or mac,i j sk i sistem (UIS) ......>
(3)	sistem podpore odločanja (SPO). Ilustrirajmo navedene usmeritve računalniško podprtih poslovnih i n-f ormac i jsk i h sistemov, kot njihove razvojne korake, s kratkim prikazom njihovih značilnosti (glej tudi sliko 1).
Vsekakor lahko ugotovimo, da gr-e za vrsto kvalitativnih sprememb, ki se kažejo predvsem v (Kajzer, 1986):
-	prenosu težišča s podatkov (AOP) na sporočila (UIS) in nato na. infarmaci je (SPO);
-	razširitvi seraiotične razsežnosi s sintakse (AOP) na semiintiko (UIS) in nato na pragmati-ko (SPO);
-	spremembi organi z i ranost i, podatkov od posamičnih (AGP) na bazo podatkov (UIS) ter nato na bazo modelov in -podatkov (SPO);
.- dopolnitvi standardnih (s)poročil z ad hoc sporočili (odgovori na vprašanjci);
-	prehodu od starejših postopkovnih (AOP) na sodobnejše postopkovno (U-IS) in nato n£i napo-stopkovne programske jezike (SPO); preoblikovanju osnove delovanja od posamičnih proyramov (AOP) na programske "pakete" (UIS) in nato na gener-atorje sistemov podpore odločanja - sisteme za reševanje problemov (SPO);
-	pr-ehodu od nei ntegr i r anost i (AOP) k integri-ranosti na ravni (poslovne) funkcije (UIS) in
nato k integriranosti na ravni organizacije (SPO);
-	povečanju stopnje prilagodlJ ivosti 'od togosti do spremenljivosti' (-fleksibilnosti);
-	razširitvi -na zgodnejše faze upravljanja: od operati ve . (AOP) na taktiko (UIS) in nabalje ' na strategijo (SPO);
-	spremembi cilja: od težnje k učinkovitosti obdelave podatkov (AOP) k učinkovitosti (UIS) in nato uspešnosti poslovnega si stema (SPO),
-	trans-f ormaci j i "uporabnika" od organizacije (AOP), na posameznika (UIS) in nato na posameznika v organizaciji (SPO);
-	razširitvi podpore strukturiran i h odločitev na podporo semi strukturi ran i h odločitev in nestruktur i rani h odi oči tev;
-	prehodu iz posamičnih modelov na sisteme modelov. ,
Vidimo, da se je na področju poslovnih in-formaci jski h sistemov is marsikaj spremenilo, mnogo pa se bo še moralo. V nadaljevanju se bomo omejili na vlogo računalnika kot pripomočka pri poslovnem odločanju. Pri tem bomo izhajali iz sistema č1 ovek-računalni k, kot lahko označimo (sinergijsko) sodelovanje človeka in računalnika v procesu računalniško podprtega poslovnega odločanja. Gre za še .enega izmed -fenomenov iz skupine, ki jih označujemo z computer aided .... (CA..) - namreč.za computer aided decision making (CADM).
2 RAČUNALNIK KOT PRIPOMOČEK PRI POSLOVNEM ODLOČANJU
Sistem človek - računalnik sestavljajo, tako kot vsak sistem, določene sestavine, notranje povezave med sestavinami in mejne povezave z okoljem.' V sistemu čl ovek-r ačunal n i k so:
(1)	sestavine izražene z množico ljudi in množico računalnikov,
(2)	notranje povezave z nekim procesom, ki določa "pravila" sodelovanja ljudi in računalni kov,
(3)	mejne povezave pa s podatki kot preslikavo stanja v objektivni stvarnosti ter z "rezultati", ki izhajajo iz sinergijskega (so)delovanja ljudi in računalnikov.
Sistem človek-računalni k lahko zato y kontekstu poslovnega odločanja razumemo kot (integriran) sistem ljudi in r-ačunal ni kov iz katerega skozi
ZnačiInosti
. teiiite
. semiotična razsežnost
. organiz. podatkov
. sporočila
programski jez i ki
osnova delovanja
AüP PODATKI
SINTAKSA
POSAMIČNI PODATKI
UIS SPOROČILA
SEMANTIKA BAZA
PODATKOV
— > SPO
INFORMACIJE
PRABMATIKA
BAZA PODATKOV IN MODELOV
STANDARDNA (S)POROČILA
AD HOC (S)POROČILA (ODGOVORI)
cilji
razsežnost upravijanja
i ntegri ranost
stopnja prilagodlj.
uporabni k
vrste odi oči tev
vrste inodel ov
STAREJSI POSTOPKOVNI
POSAMIČNI PROGRAMI
UČINKOVITOST OBDELAVE POD.
OPERATIVA NE OBSTAJA
TOG
ORGANIZACIJA
STRUKTURIRANE S
POSAMIČ
NOVEJŠI POSTOPKOVNI
PROGRAMSKI "PAKETI"
UČINKOVITOST ORGANIZACIJE
TAKTIKA
NA RAVNI FUNKCIJ CE)
OMEJENO PRILAGODLJIV
POSAMEZNIK
E M N
NEPOSTOPKOVNI
SISTEM ZA REŠEVANJE PROBLEMOV
USPESNOST ORGANIZACIJE
STRATEGIJA
ORGANIZACIJA KOT CELOTA
FLEKSIBILEN
POSAMENIK V ORGANIZACIJI
I S T R U K
I
S I
MODEL S T E M I
T U R I R A N E NESTRUKTURIRANE I
MODELOV
Slika 1: Razvojni koraki modelov poslovnih informacijskih sii temov (Kajzer, 1986).
(računalniško podprto) poslovno odločanje "izhajajo" poslovne odločitve. Proces, ki poveže ljudi in računalnike je torej v tem primeru odločitveni proces, rezultati pa so poslovne odločitve.
Ce lahko računalniško nepodprto poslovno odločanje ponazorimo takole
človek —> odločitev oziroma tudi takole, ko človek sicer uporablja računalnik, vendar ne (neposredno) pri odločanju.
računalnik —> človek —> odločitev za razliko od računalniško programiranega odločanja, ki poteka takole,
človek —> algoritem —> računalnik —> odločitev,
potem lahko računalniško podprto odločanje ponazorimo takole človek
I --------> odločitev.
računalnik .
človek
od 1 oči tveni proces
I računalnik !
Slika 2: Sistem človek - raČLinal ni k pri poslovnem odločanju.
t;;oncept sistema človek ■■ računalnik izhaja iz predpostavke, da so nekatere naloge najlažje izvedljive, če se jih loti človek, druge pa, če se jih "loti" računalnik. V odvi-
janju poslovnega odločanja oblikujeta človek in računalnik ( i n-f ormaci jska tehnika in tehnologija) sistem, iz katerega izhajajo rezultati (poslovne odločitve) in sicer skozi vrsto interakcij med računalnikom in človekom (glej sliko 2).
Izhodišča sodobnega računalniško podprtega poslovnega odločanja, ki se odvija v tovrstnem sistemu človek-računalnik, izhajajo iz naslednjih predpostavk o vlogi računalnika pri tem :
(1)	Računalnik naj nudi podporo odločevalcem pri poslovnem odločanju, torej jih naj le podpira in ne nadomešča oziroma zamenjuje. Predvsem presoja ljudi je tista, ki je načeloma nikoli ne prepustimo računalniku (Keen, 1976).
(2)	Poslovne odločitve naj nastajajo v neposrednem dialogu človek-računalnik, tako da računalnik nudi pomoč pri razumevanju konteksta odločanja, omogoča različne pripomočke za računalniško podporo odločitvenih procesov (operacij), zagotavlja dodatni, z vidika človeka odločevalca eksterni spomin ter dopušča učinkovito upravljanje računalnika, kot pripomočka poslovnega odločanja (Sprague, 1983).
2e sredini 50-ih let je Rober t Oppenheimer (Chora-fas, 1986) ugotavljal, da je računalnik stroj, ki je popolnoma drugačen od elektronskega knjižnega stroja in od ostalih "strojev" za obdelavo podatkov. Zato ga moraino uporabljati nil popolnciiria drugačen način, če hočemo kar najučinkoviteje izkoristiti vse prednosti, ki nam jih nudi. Potrebovali smo tri desetletja, da smo doumeli bistvo te misli. Nadoinestiti moramo
zamujeno in narediti bistven korak naprej.
Raiunalnik moraino začeti uporabljati predvsem kot procesor "simbolov" in ne le kot procesor števil tkal kulator ) . Uporabiino ga za učinkovito računalniško podproro t.i. kognitivnih procesov. Le tako bo računalnik lahko resnično postal pripomoček človeku - posamezniku.. To nam danes uspeva različno (ne)uspeèno.
Računalnike moramo začeti uporabljati tudi za pavesovanje posameznikov, ki odločajo v vzpostavljeni organizacijski strukturi poslovnega sistema. Posamezniki, ki' odločajo niso le "solo èahisti", ki "igrajo" - odločajo samostojno, temveč odločajo predvsem v skupinah. Zato moramo začeti razmišljati tudi o tem, kako računalniško podpreti skupinsko odločanje (DeSanctis, 1984) ter kako računalniško podpreti komuni C i ran je med posainezniki (Winograd 1986).
.3 ALI LAHKO RACUNf^LNIK ODLOČA IN MISLI ?
Menimo, da lahko t-ačunalnik odloča le v primerih, ko zadošča obnavljanje vnaprej pripravljenega postopka. V teh primerih lahko govorimo o programiranem odločanju (Simon, 19t0). Odločanje lahko programi ramo le takrat, kadar lahko opredelimo natančna pravila,, po katerih se bo odločanje odvijalo. Ta pravila so osnova za računalniški program, ki bo "prevzel" odločanje in s tem nadomestil človekovo presojo. Računalnik pa v tem primeru ne bo mislil, bo le obnavljal, kar pomeni v besednjaku kognitivne -psihologi je učenje in ne mišljenje.
Zaključimo naše razmišljanje s tem, da gre pri vprašanjih, ali lahko računalnik odloča in ali lahko računalnik misli, za dve različni stvari. Računalnik lahko odloča, vendar ne na način, ki ga označujemo kot miselni proces - mišljenje.
4 MODELI RAČUNALNIŠKE PODPORE POSLOVNEGA ODLOČANJA
Kljub vsemu pa ne smemo pozabiti, da je računalnik le stroj in zato od njega vseeno ne smemo pričakovati preveč (Kajzer, 1987). Kakor vemo, je računalnik še daleč od vsemogočnega "mislečega" stroja, ki bi lahko zamenjal človeka, vendar pa ga nikakor tudi ne smemo podcenjevati: na številnih področjih človeka odi oči 1 no -prekaša in je (lahko) nenadomestljivo orodje. Kako smotrno ga bomo znali uporabiti, pa je odvisno od nas samih. Nevarno je, na žalost zelo razširjeno mnenje, da lahko vse zmogljivejša računalniška tehnika in tehnologija zamenja človeka ne le pri. rutinskem, temveč tudi pri ustvarjalnem delu. Kakor ugotavlja K-iaj-zer (Kajzer, 19S7): "Ce le nekoliko pretiravamo, bi to poiiianilo, da bodo vse bolj "butasti" ljudje z vse "pametnejšimi" računalniki čedalje ustvar-jal ne j ži . Kakšna .zabloda 1 V resnici je položaj ravno obraten: čim boljše in učinkovitejše je orodje, tem več mora biti. v gl. avi. in roki, ki g'a uporabljata."
ro začetnih naporih "umetne inteligence" in računalniku kot "mislečem" stroju (Moto-Oka,I 1984) , se je tovrstni zanos nekolika umiril, kajti vse bolj je jasno, da je računalnik le stroj in bo to vedno tudi ostal. Odgovoru na to vpriišanje se je (tiogoče približati, če razmislimo, kaj pravzaprav pomeni pojem inteligenca. Ta uganka vzemirja -filozofe ža od nekdaj in resnično je ni lahko raz voz 1 i-i t i . Jasno je, da lahko inteligenco obravnavamo z več vidikov, npr. kot sposobnost načrtovanja, učenja, reševanja problemov, uporabe in razumevanja jezikov itd. Obnašanje, ki kaže (vsaj) eno izmed teh sposobnosti, šteje večina ljudi za inteligentno. Cilj tistihi ki se ukvar-jajo z umetno inteligenco, je torej pripraviti računalnike do tega, da se bodo sposobni tako obnašati. Zato so tudi izhodišča ekspertnih sistemov, kot "nadomestka" umetne inteligence postavljena nekoliko drugače (Schnupp, 19S5). Izhajajo iz predpostavke, da lahko razvijemo sisteme čl ovek-r-ačunal ni k , v katerih bi se računalnik obnašal "inteligentno", vendar ne na takšen način kot to počne človek.
Računalniško podporo poslovnega odločanja je vedno bila in bo tudi v bodoče predvsem heterogena celota, nikakor- pa ne homogena oziroma monolitna "tvor-ba", čeprav bi to mnogi želeli. Zaradi tega jo moramo na različne načine organi z i rati. Predvsem pa moramo spoznati, da računalniška podpora odločanja nikakor ni homogena celota, kot smo še donedavnega mislili. Menimo, da jo moramo, v kolikor hočemo ponazoriti omenjeno heterogenost, ter ugotoviti, kje smo in kam gremo, analizirati vsaj z vidikov:
(1)	metod odločanja, ki jih želimo računalniško
podpreti (analitične metode -i---> hevris-
tične metode)j
(2)	stopnje presoje, ki jo prepustimo človeku (malo človekove presoje -C---> veliko človekove presoje);
(3)	števila posameznikov, ki sodelujejo ptn
(računal niško podprtem) poslovnem odločanju (individualno poslovno odločanje -f.---> skupinsko poslovno odločanje). Ob upoštevčinju omenjenih treh vidikov, lahko torej ločimo naslednjih pet tipov računalniške podpore poslovnega odločanja, ki se med seboj, čeprav služijo istemu namenu, v mnogočem razlikujejo:
(1)	Računalniška podpora tip I podpit-a analitične metode, ne dopušča veliko človekove pr-esoje in je namenjena podpori individualnega poslovnega odločanja. Gre za računalniško podporo odločanja, ki jo pogosto označimo kot strukturirano poslovno odločanje (angl. SDS) - odločevalec izbere model, po katerem bo računalnik odločal (Börry, 1976).
(2)	Računalniška podpora tip II podpira analitične metode, dopušča veliko človekove pre-
■soje in Je namenjena podpori indivdualnega poslovnega odločanja. Tovrstno računalniško podporo poslovnega odloč.anja najpogosteje imenujemo sistem za podporo odločanja (angl. DSS). Opredelimo jo lahko kot tisto, ki odločevalca soočenega s semi struktur i rano odločitvijo, podpira skozi dialog člo-vek-računalni k, možnostjo učinkovite uporabe podatkov in vrste modelov za analizo
problemov ter sintezo odločitev (Spragu.e 1983).
(3)	Računalnižka podpora tip III podpira hev-ristične metode, ne dopu.ida veliko človekove presoje in je namenjena podpori individualnega poslovnega odločanja. Danes najbolj razširjen predstavnik tovrstne računalniške podpore so ekspertni sistemi (angl. ES), ki odločevalcu sugerirajo odločitev, oblikovano na podlagi v računalniku strukturi ranega (strokovnega) znanja in posplošenih, problemsko neodvisnih hevris-tičnih metod (Kr^allman 1987).
(4)	Računalniška podpora tip IV podpira hevris-tične metode, dopušča veliko človekove presoje in je namenjena podpori indivi dui nega poslovnega odločanja. Gre za ekspertne sisteme za podporo odločanja (angl. ESS), ki za razliko od ES dopuščajo izbirno hevris-tične metode (Luconi 1986)..
(5)	Računalniška podpora tip V je namenjena podpori medsebojno (so)odvi snega poslovnega odločanja posameznikov, ki sodelujejo, najpogosteje skozi semi struktur i rano poslovno odločanje kot skupina (angl. SDSS) (DeSan-ctis, 1985).
Večina omenjenih modelov računalniške podpot-e poslovnega odločanja je sicer raziskanih, z ostalimi pa se in-formatika kot znanost intenzivno ukvarja. Menimo pa, da je kar (pre)pogosto kori-tekst njihove uporabe bliiji upravljalnemu informacijskem sistemu (UIS), kot pa omenjenemu sistemu podpote odločanja SPQ (glej točko 1). Zato tudi menimo, da jih moramo (v nekaterh primerih) postaviti na prava izhodišča, predvsem pa poiskati njihove stične točke in ti^ko ugotoviti, kako jih povezati v heterogeno celoto, ki bo omogočala uspešno in učinkovito celovito podporo poslovnega odločanja.
Računalniška podpora tip I je "klasična", najstarejša, zato je na tem področju tudi relativno največ narejenega. Zakaj relativno ? Zaradi tega, kef je sicer- resnično veliko narejenega, vendar je ravno tukaj kontekst uporabe računalnika največkrat sporen. Relativno nekaj manj je narejenega pri računalniški podpori tip II. Zakaj prav tako relativno ? Veliko vprašanj je namreč še odprtih, čeprav je koncept uporabe računalnika najpogosteje pravi. 3e manj je narejenega na področju računalniške podpore tip III. Na ostalih področjih (tip IV in V) pa je narejenega resnično zelo malo.
Vemo, da se tako sistemi za podporo odločanja (tip ID, kot podobno tudi ekspertni sistemi (tip III), nekako razvijajo v ekspertne sisteme za podporo odločanja, (tip IV). Prav tako pa se predvsem sistemi za podporo odločanja (tip II) usmerjajo na področje računalniške podpore skupinskega odločanja (tip V).
5 IN KAKO NAPREJ ?
Ker je poslovni sistem mnoiica na določen način med seboj povezanih posameznikov, oziroma mnoiica skupin posameznikov, ki sodelujejo z namenom doseganja postavljenih (skupnih) ciljev, je model i n-f ormaci jskega sistema, ki smo ga označili kot sistem podpore odločanja
(SPO), z vidika računalniške podpore dejansko množica medseboj povezanih sistemov človek-ra-čunalnik. V tem primeru torej lahko (in moramo) govoriti o nekakšni mreži sistemov človek-raču-nalni k.
Vspostavitev sistema podpore odločanja (SPO), kot mreže sistemov človek-računalni k naj upošteva tako kategorijo odnosov človek-človek, kot tudi kategorijo odnosov človek-računalni k ter kategorijo odnosov računalnik - računalnik. Vse odnose moramo opazovati v kontekstu organizacije. Menimo, da so predvsem prvo, pa tudi drugo omenjeni odnosi predvsem (nikakor pa ne izključno) problem (ekonomske) in-formatike, zadnje omenjeni kot tudi drugo omenjeni odnosi pa predvsem (vendar tudi nikakor ne izključno) problem računalništva. Vsekakor smo postavljeni pred velik izziv.
LITERATURA:
/1/ Chora-fas D.; Fourth and Fi-fth Generation Programming languages, McGìraw-Hi 11 , New York, 1986.
/2/ DeSanctis G., Ballupe: Group DSS: A New ■frontier, Data Base, 1985 Winter, str. 3-10.
/3/ DeSanctis G. , Gallupe: In-formation systems support -for group decision making, MISRC-WP-85-115, Univ. Minessota, 1984. /4/ Gorry,G.A., Scott-Morton M.S: A Framework ■for MIS, Sloan Management Revitsw, 1976/1, str. 55-70.
/S/ K.ajzer S. , Marn F.: Cesa računalnik v poslovnem sistemu ne zmore, Organiziranje in razvijanje in^formaci jskih sistemov, Radenci, Društvo ekonomistov Maribor, Maribor, 1986.
/6/ Kajzer S. in sodelavci: Raziskava izhodišč za oblikovanje modela poslovnih informacijskih sistemov, VEKS Maribor 1986. /7/ Keen P.G.W: Interactive computer systems for iiiiinagers; A modest proposal, Sloan Management Review, 1976/3. /S/ Krai Iman H.: Betriebliche Entscheindungs-Unterstutungsysteme. Zeitschrift Fuhrung + Qr^ganisation, 1987/2, str. 109-117. /9/ Luconi F.L., Malone T.W., Scott-Morton
M.S.: EKpert Systems:The Next Chalenge for mangers, Sloan Management Review, 1987/4, str. 3-15.
/10/ Moto-Oka T., Kitsuregawa M.: The Fifth
Generation Computer, Wiley, New York, 19B4. /11/ Schnupp P. in Leinbrandt U.: Expertsn-systeme, Nicht nur fur Inf ormati ker, Springer, Berlin, 1985. /12/ Simon H.A.: The New Science of Management
Decisions, Harper Row, New York, 1960. /13/ Sprague R.H., Carlson : Building effective DSS, Prentice Hall, Eng-lewood Cliffs, 1983. /14/ Winograd T., Flores; Understanding Computers and Cognition, Ablex Pubi. Corp., Norwood, New Jersey, 1986.
UUCP MREŽA
INFORMATICA 4/1988
UDK 519.876 UUCP
Boris Domajnko Iskra Delta
UUCP nelioori:; is a system, mai lii'ìg, i" eni Coni or mafc i an it assi.u-e;s -futur e. Cither-accepted sta presented in ISO/OS I- model tools will them are ment
software- is a |:iart o-f U compound o-f -l-iles arid cite , >::;:i::)amiand e;< eciut i'::jn
w i t 0 SI' r e -f e r enee !(i a d e 1 a(n(::mc3 other
IIX distribution, uucp ( UniK to IJni;-; Copy ) programs, used -for network i 1 e trans-fer, and direct access to adjafrent computer, is important for each computer network
::|ualitieB a possibility of f urther ' i mprovements in the wise' its .isage will cease and other'■ pr o tocol s with internationally ndards wi-11 be implemented instead. ■ A short analysis of uucp -is' this paper with an aye on the OSI model known in the t i nie wlien i-u.>cp was created, till be used for some ti ine because o-f ecJ : amer:i. can LjlJOP, euro|::ieari EUnet
though there was no But uucp communication e i -s t i n g n e t wo r k s. T h r e e o-f and Yugoslav ZtlUTA.
Del -standardne d i str" i t'Lic i j e LIN'tXa je uucp mreìa. uucp (^Unii< to Uni;-; Copy ì ' je Ì2 datotek in programov sestavljen sist63tn, ki omogoia u'pCir-€j.bo uka;:ov' za prenos datotek in elektronske poèta med računalniki, izvajanje ukaiov na oddaljenem r-aćunal n i t-::u ter- direktna pr;-i k 1 j uiü; i tev na sosecJnji računalnik. Ustre-;:an je OSI referenčnemu modelu daje. vsaki računalniški mreii poleg ostalih kvalitet tudi možnosti ■ za nadal jni razvoj. V naspr-otnem primeru jo. ba slej .ko prej ^ potre-ibno zamenjati s , sodobna^ j-ži m konceptom. ' V času. nastanka,- üucp-j'a ISO/OSI', referenčni model še.ni 'bil znan." KI jub, no v i ni reèitvam pa bo Se nekaj časa ■ moina njegova uporaba zaradi že obstoječih mr-ei. Omenjene so tri: .UUCP, GUnet inZUUTA .
1. Uvod
Unix je operacijski sistem, ki se ie razvija v obliki novih'verzij in- z ahte?vne j-=i h standardov, npr. AT?/T System V in POS IX -- IEEE standard 'v fazi raz-voja. Mreže -so bile dolgo 'časa največji problem tega operaci j'skega sistema, saj je UNIX éele s kasnejšimi verzijami postal -mačno olio.lje za razvoj in izvajanje mrežnih aplikacij. Se iz prvih časov obstaja uucp mrež.a, ki je del standardne di str i taue i-j e LÌMIXa. Oglejmo si .jo' zaradi rijene pr epr osto-Bt i , r-az éi. r-j eru:;ist i in možnosti takoj^inje uporabe.
2. uucp sistem
uucp ' ( Unix to unii«; CoPy ) je' sistem, sestavljen iz datotek i. ri pr-ograinov, k:i omogoča prenos datotek'in elektronske poéte med dvema r-ačunal n i koma, izvajanje t.,(kazo%'' na oddal jeriefii računalniku. ter direktno priključitev na sosednji računalriik. Spada v sl-:ui::) i no fiir-ež , kii so globalne glede na določen operacijski sistem (2) .
Minima-lna konf i gur ac i ja z-a uucp mrežo je sesta v 1 j B n « i z d v e h r a č u n a 1 n i l< o v, ki j u povezuje serijski kanal V.24 ( ämer. RS-232C i.
Za vzpostavitev zveze sta" možna dva načina. Pri prvem je naé računi".! ri i k gcrst. i. tel j in ':::imogoč:a., da se nanj priključi. oddaljeni računalnik. Drugi način se izvaja takrat, ko naš računalnik kliče oddaljeni računalnik ( npr. preko javnega t e 1 e f o n s k e g .a -'.a li. p a t; e t n e g a o m r- e ž j a ) .
asinhroni en način ser i'jski klic iz ko hočemo
Kadar ■ za povezavo uporabljamo ■serijski kanal, l-ahko iz.va.jamo samo naenkrat. Ciä upor-ab 1 jamo samo en kanal, je običajno na vol-jo za oddaljenega računalni ka.- 'V primeru, k;]. :i. cat. i ve^n , začasno pre-'povemo dospele, k: lice in si" r ezié.r vi ramo kanal. Slednje velja le, če u p o r -a b 1 j a m o m o d e m.
3. Pripr.aya uucp sistema in opis delovanja
I m e u u. C p je bil o v z a č e t k: u povezano s transportnim nivojem pri povezavi UNIX oz. sorodnih računal ni kov. Mreža deluje; tako, da sosed'. predaja sporočilo sosedu, dokler sporočilo ne pride do končnega naslovnika, uucp je ,-ena od redkih mrež, kjer izvor določi .pot podatk:ov. Pri večini ostalih mrež 'je usmerjanje dina.fni čno i rt ne fiksi'io.
Kot že rečeno, je uucp -si stein datotek in programov. Da.toteke vsebujejo informacije o mreži oz'. o vlogi na-ž;ega računalnika v njej, prav tako pa tudi informacije o podatkih, ki j i l~i o č e m o p r e n a é a t i . D a t o t e k: a ■ m o r amo imeti .ločeno za oba nači-na -■ Iclicni in klicani. -Oglejmo si nekaj, bistvenih datotek, da osvetlimo informacije, fci jih 'ima računalnik v zvezi z uucp mrežo.
Batoteka. /e-tc/systemid hrani ime kraj I in ime naéega računalnika.
lokacije (
V datoteki up or a. bri i k i. gesla. Poleg
/etc/passMd
rači...(i"ial ni k a i n tega je -tu še
so opisani vsi n j i h o v a. k o d i r a n a ■ ime programa, ki naj
se zaćne izvajati, ko uporabnik pove svoje ime ( login ). Za h:l.ice in mreie torej tu zaznamujemo avtomatičen start programa, ki bo bedel nad nj imi.
Datoteka	/usr/1ib/uucp/USERFILE	vsebuje
omejitve v zvezi z zunanjim dostopom do našega računalnika, npr. do katerih direktorijev ima določen uporabnik iz mreže dostop.
Za pomoč pri dostopih do drugih računalni kov pa nam sluii datoteka /usr/1ib/uucp/L.sys, ki
vsebuje podatke o računalni ki h, ki jih lahko kličemo, npr.: ime računalnika, dneve z dovoljenim dnevnim časom dostopa, vrsta in étevilka kanala, hitrost, tele-fansko številko ter določene informacije za postopek prijavljanja na oddaljeni računalnik.
Za jasnost podatkov v prej omenjeni datoteki imamo še datoteko /usr/1ib/uucp/L-dialcodes ,
ki vsebuje okrajšave za določene telefonske étevilke, ki jih pogosto uporabljamo.
Poglejmo še najvažnejše programe v sistemu uucp.
Ce hočemo prenašati datoteke, jih prepustimo programu uucico < Unix to Unii-; Copy In Copy Out ). Le-ta izvrši delovne datoteke za prenos podatkov, ki jih je pripravil uucp. Ima pa tudi glavno vlogo pri prenosu v obratno smer, torej pri sprejemu.	Klic običajno izvajamo
periodično in avtomatična, tako da definiramo v datoteki /usr/1ib/crontab, kdaj naj poseben sistemski program kliče uucico. Ta ob izvajanju pogleda, če ima na določenem direktoriju zahteve za prenos. Ce obstajajo, izvrši prenose. Prenos se lahko izvede tudi v obratni smeri, če na oddaljenem računalniku čakajo zahteve za prenos na našega. Uporabniku tako ni treba skrbeti za vzpostavitev zveze, .zadošča mu, da ve, da se prenos izvaja npr. vsake pol ure ali pa npr. v nočnih urah.
Program uucp poskrbi za prenos datotek, tako da najprej kreira določene delovne in podatkovne datoteke. Ce prenos ni možen takoj, ga bo uucico poskušal izvesti v časovnih intervalih, ki so definirani v datoteki crcintab.
izvrši ukaz na določenem računalniku. Datoteke, ki so povezane z ukazom, lahko prečita' na različnih računalnikih. Poglejmo za primer, kako primerjamo dve datoteki z imenom dat. Prva se nahaja na. računalniku ti, druga na računalniku t2, rezultat pa naj bo shranjen na našem računalniku v datoteki rez. Običajna ukazna vrstica bi bila ta.ka-le:
diff /usr/a/dat /usr/b/dat > res .
Ukazna vrstica z uux ukazom je podobna običajni, le da dodamo ukaz uuK ter definiramo imena! računa1ni kov, kjer se posamezne datoteke nahajajo:
"diff tt!/usr/«fdat 12 !/usr/b/d at
' rez "
uuxqt izvrši ukaz, ki pride iz oddaljenega računalnika. Preveri, če je u.ka:'. dovoljen, saj posameznim zunanjim uporabnikom lahko zelo omejimo pravice glede klicanja programc.iv na našem računalniku.
■ail je sistem za elektronsko pošto. Obsega pripravo sporočil in pošiljanje oz. sprejemanje sporočil. Uporablja uucp sistem.
cu je program, ki nam omogoča, da. se k:ot terminal priključimo na sosednji računalnik.
4. uucp in OSI referenčni model
Protokoli v računalniških mrežah so lahko zelo kompleksni. Da jih lažje obvladamo, so izdelani v različnih nivojih. Spodnji so bolj povezani s strojno opremo, najvišji pa predstavi jajo stik z uporabnikom. Nivoji so definirani ob takšnih zahtevah (7);
-	nov nivo je kreiran tam, kjer je potreben nov nivo abstrakcije,
vsak nivo mora izvajati dobro definirane funkci je,
-	funkcije vsakega nivoja morajo biti izbrane v smislu definiranja oz. mednarodna definiranih protokolov,
-	meje nivojes' morajo biti izbrane tako, da je pretok informacij med njimi minimalen,
-	število nivojev mora biti dovolj veliko, da ni treba mešati določenih funkcij, in dovolj malo, da celoten sistem ne postane preveč zapleten.
ÜNIXov uucp protokol je bil izdelan za povezave, ki niso stalne. V času njegove definicije OSI referenčni model še ni bil znan in če smo zdaj za implementacijo ISO standardov, moramo sprejeti uporabo novih protokolov (8). V začetni fazi pa je še vedno potrebna možnost uporabe standardnih UNIXovih komunikacijskih orodij ( cu, uucp ) in aplikacij ( npr. mail ) preko asinhrone ali X.25 linije. To je nujno potrebno že zaradi nemotenega delovanja obstoječih mrež, dokler se le-te ne prilagodijo standardom.
Programska oprema za uucp je sestavljena iz več delov. En del služi za pripravo zahtev za prenos, drug del nastavi karakteristike kanala in vzpostavi zvezo ob uporabi pravilnega gesla, na koncu pa tudi zaključi zvezo. Tretji del pa se izvaja mc^d prenosom in pozna dva načina: klicni in klicani. Poskrbi za oddajo in sprejem datotek, prav tako zna sprejete podatke poslati naprej proti končnemu naslovniku. Uporabljen je lasten protokol s kontrolo pravilnosti podatkov. Ce je spodnji nivo zanesljiv ( npr. X.25 ), to kontrolo izp.jstimo.
Naredimo zdaj poskus podreditve funkcij uucp-Ja ISO/OSI modelu, kar ni lahko, saj le s težava ločimo funkcije 5., 4., in 3. nivoja.
7. APLIKACIJSKI NIVO
Pravzaprav la.hko celoten uucp sistem imenujemo aplikacijski program, saj omogoča ( tudi drugim programom ) prenos datotek in elektronske pošte ter priključitev na sosednji računalnik.
6. PREDSTAVITVENI NIVO
Ker- uucp ni mišljen za sodelovanje heterogenih sistemov, ne oniogoča prilagajanja lokalne sintakse, razen pri prenašanju neznakovnih datotek. Ostali programi za obdelavo datotek pred oddajo oz. po sprejetju niso del uucp sistema in jih izvajamo posebej ( npr. kriptografska obdelava ). Pri oddaji pa vseeno lahko ustrezne ukaze definiramo na istem mestu kot ostale ukaze za pošiljanje.
5. NIVO SEJE
Nivo seje nudi funkcije za povezavo dveh api i kaciJskih procesov v odnos logične komunikacije in funkcije za organizacijo in sinhronizacijo njunega dialoga. Preden dva aplikacijska procesa začneta komunikacijo,
morata vzpostaviti, komunikacijo n« nivoju seje. Pri uucp-ju je ta pogovor dvE?h proigramov uucico, ki se obić-ajno ne izvaja takoj ob zahtevi za prenos , ampak ob določenih časovnih intervalih. Tako lahko pozabimo n-a dvosmerno soiaano razpravo med dvema .aplikacijskima procesoma. Pomemtana vlogo im-rsjo i n-f or mac i je , ki so shranjene v r.azličnih datotekah;
-	dovoljenja za dostope iz drugih r ačuna.l n i kov,
-	dovoljenja z.a i zvrševanje ukazov iz mreie in seznam dovoljenih i.kazov ( za vsakega zunanjega uporabnika posebej ),
-	datoteke s podatki o povezav.ah z drugimi računalniki in datoteka okrajiav imen.
Povezava bo razpu.žčena, če ne dobimo odgovora v določenem času. Ves potek dialoga se zapisuje v dnevnik, t.j. v posebno datoteko.
4. TRflMBPORl'MI NIVO
Pri uucp-ju ni pr-ave meje med transportnim in mrežnimi ' nivoji, detajli transportnega protokola pa so poleg vsega ■■je nedokumentirani. Ta nivo je n^a voljo tudi za pošiljanje elektronske po^ète ( "m^ail"), ki je vezano na uucp sistem. Na splo#,no za 'poétni del velja, da je narejen eksplicitno z aplikacijskimi protokoli,	implementi rani mi z ' uporabo
izvrševanja ukiazov na oddaljenem rači.in-al n i ku.
Transportni nivo po pravilih dobi podatke iz nivoja seje in jih oblikuje v pakete. Ker gre pri uucp-ju pravzapr-av za isti program, nivoja v zvezi 3 čistimi podatki težko ločimo. Ko preda podatke mrs/žnemu nivoju < krmilnemu programu za serijske linije ), čaka potrditev od prejemnika. Vključena- je tudi kontrola prenosa. .V primeru uporabe X.25 protokola kontrole pravilnosti pod-atkov na tem nivoju ni zaradi hitrejšega prenosa. Transportni nivo odigra svojo funkcijo tudi v primc-?ru, ko ugotovi, da ni končni naslovnik in rapra zato prejete podatke poslati naprej.
Spodnji trije nivoji so odvisni od metode prenosa podatkov v mreži ( navadna serijska linija, V.21: modemska povezava, X ; 25, LAN).
integr.acijo določenih proi z vajal cev s ciljem, da združijo moči pri pr i 1-agod i t vi LINIXa OSI nivojem in ■ s tem omogočijo priključitev na d r u g a č n e r a č u. n a 1 n i k e in mreže, če le u s t r e z a j o mednarodnim standardom, so predvideli uucp samo 'iie v vmesni + az i svojega projekta. Z uporabo uucp-j^a preko X.2S5 in LAN mreže so se izognili klasičnemu problemu pri imp1ementaci j i OSI protokolov, t,j. pomanjkanju ustreznih servisov v obdobju, ko ée ni i (lip 1 emen t i r an protokol na vseh nivojih.
elek;tronska pošta., novice, p r^ en os datotek ■	
ui.tcp	
ISO transportni razred 2/3	ISO transportni razred 4
X. 25	
	LAN CSMA/CD
y n.y.slednji -fazi pa se uucp izgubi med -funkcijami, ki jih natančno določa ISO standard. Temeljito izdelan -ISO nivo seje in posebna knjižnica potrebnih -funkcij omogočata kasnejžo izdelavo drugih aplikacij, ne samo obcjh spodaj omenjenih. Mreže posameznih družb, ki ne podpirajo OSI model a ,■ ■se bodo priključile preko posebnih računalnikov ( Gateway ), posamezni PTT servisi pa do tistih OSI protokolov, ki' jih podpirajo.

:)ostop do datotek ali njihov prenos ( ISO FTAtI elektronska poèta in kon-ferenca ( X.4i<x )
ISO nivo seje ( ISO 8.326/7 ) ( -funkcije v. jedru operac i jskega sistema )
ISO transportni razred 0/2/-3
3,2,1
X.25
ISO transportni razred 4
LAN CSriA/CD
Gateway
3. MRE2NI NIVO
Identiteta končneg in-f ormaci ja , saj d i nam i čnega usmer j a določa pot podat'-: njihovi poti pa vs svoji lastni presoj ■f i k:sno usmer j ar, j -3 vse povezave. Račun najpri marnejšo pot treba vnašati loč sestavljanju ba.re računal ni ki,. z-a ce:;' bolj pogosto, saj kon-f i guraci j i cel o t
naslovnika ni zadostna ne uporabljamo običajneg nja, ampak -fiksno. Izvor sam OS',; transportni nivoji na eeno imajo možnost, da po ta pat spremi nja jo. T.a !::■:= no potrebuje bazo s podatki za ^l.nik saai iz te l::)aze ugotovi I tako da je uporabniku ni la. -Problem je. v zamudnem , ki se izvaja na enem ;itno mrežo in mora biti čim norardc si.sditi spremembam v ie mr-e^e.
Tak bi bil opis -standar dnega uucp-j à. Na preprostem modelu lahko mesto prvotnega uucp-ja pri kažemo tako:
UUCP mreža v ZDA
UUCP je ime ene od velikih nacionalnih mrež v ZDA. Mreža spada med "mreže uporabnikov s podobnimi interesi". Njeni prvi začetki segajo v leto 1978. Povez-avci geogra-fsko oddaljenih računalnikov v večini primerov še zdaj omogoča tele-fonsko omrežje. Prenos datotek in izvrševanje ukazov na oddaljenem računalniku sta bila osnovni name^^n take povezave, možnost poéiljèinja elektronske pošte pa se je pojavila kasneje (3). To možnost so lahko začeli uporabljati tudi računalniki z drugačnimi operacijskimi sistemi. Prilagajanje sintakse imen zahtevam ARPA Internet mreže kaže na interes za boljšo povezaveoz ostalimi mrežami, daje pa tudi^ možnosti za izvedbo- boljšega usmerjanja.
Tak uucp, k-3kršen Je prikazan na gornjem modelu, z OSI modelom pravzaprav nima zveze. Člani projekta ROSE (S)., ki predstavlja
C e n t r a 1 n a a d m i n i s t r a c i j a v U U C: P obstaja. Stanje bi lahko primerjali
mreži ne ari ar h i j o.
E-iolje so organizirani manjši deli mreže, ki jih povezuje večji medsebojni promet. Za priključitev na mrežo je sicer dovolj, da najd'emo sosed-a, ki je že v mreži in ki se strinja, da bo naš sosed. Svoj naslov še vnesemo v poseben seznam, da nas bodo lahko našli tudi drugi računalniki. Ce imama svojo lokalna mrežo, dodelimo računalnikom znotraj nje večje pravice, enega pa določimo za povezavo z ostalim delom UUCP mreže. Tako nam
ni treba podvajati podatkov a zunanji mreii, ker pa imajo raiiunalniki iz zunanje mreie dostop v nažo le prueko enega, na ene-Mii mestu laije skrbimo za i;-vajanje varnostnih aspektov.
Vsak računalnik: plačuje zs. svoje? pover^ave. Zato nekateri podpirajo priključitev čim večih sosedov, saj si s tem zagotovijo krajéo pot za lastna sporočila. Stroéki mrceže so v glavnem povezani s plačilom telefonskih storitev.
Ce upoétevamo sorodne mreie, s katerimi je povezan, je v UUCP mreiah medsebojno povezanih med 7000 do 100013 računalnikov (3).
6. Evropska UNIX mreža
EUnet
Iz UUCP mreie se je leta 1980 razvila v ZDA že USENET, ki omogoča med drugim tudi X.25 povezave. Leta 1982 pa je zdruienje EULIG ( European Unii< systems User's Group i ustanovilo evropska mreio EiUnet ( "European UNIX net" ). Ker utegne biti zanimiva tudi nas, si jo oglejmo.
EUnet ima za vzor UUCP in USEt'JET iz Severne Amerike.	Večina računalnikov ima UNIX
operacijski sistem. Organizacija mreže je boljia kot v Severni fimeriki, saj za njo stoji organizacija EUUG. Mreža je neposredna ali posredno priključena na ostale evropske in neevropske mreže. Njen glavni namen je poéiljanje elektronske poète in novic (4), preko X.25 pa je možna tudi dir^ektna pr i ki ji.iči tev ria oddal jni računalni k.
V vsaki državi je po e-M't gl avrti ra.čurtal n i k., ki skrbi za mrežo v svoji držitvi , včasih kar z direktno povezavo z vsemi ostalimi računalniki. Glavni računalniki posameznih držav lahko komunicirajo med seboj. Obstaja že glavni računalnik za celo mrežo, ki povezuje glavne računalnike posameznih držav - to je računalnik ■cvax v Amsterdami..A. Ta rai-:unalni k je povezan z ostalimi UNIX mrežami : ACSnet < Avstralija ), USENET ( ZDA ), CDAnet ( Kanada >, JUNET ( Japonska ), SDN ( Koreja ) in že mreži v Izraelu in Novi Zelandiji. mcva;; pa ima povezave že z mnogimi drugimi mrežami, ki ne bazirajo na UNIX operaci jsl::em sistemu. Prehcad na druge mreže lahko nudijo tudi nekateri drugi računalniki v mreži.
Ta mreža se žiri. Najprej so jo uporabljali entuziasti, ki so si požiljali sporočila in imeli na ta način skupne ci£?bate. Sedaj je krog uporabnikov dosti večji. To so univerze, posamezna podjetja ter reklamna podjetja. Povezuje jih tudi EUUB združenje, ki svojim članom ( uporabnilsom ÜNIXa ) tudi omogoča dostop do različne p r o g r a ms k e op r e m e. 1z d a j a j o svojo revijo, ki je vsa namenjena LINIXu. V okviru združenja tečejo tudi razgovori o prilagoditvi	ISO/OSI modelu na nivoju
komunikacije med meddržavnimi vozližči.
V Evropi je priključenih približno 80(3 računal n ikov iz 16. držav.
System <IES). Trenutno se ukvarjajo s prototipom OSI okolja pod UNIX System V in Berkeley 4.2BSD. Sicer pa zaenkrat pripada večja pozornost skupini protokolov, ki v spložnem ustreza 3. in 4. nivoju ISO/OSI referenčnega modela, čeprav je bil slednji narejen pozneje. To je TCP/IP ( Transmission Control Protocol / Internetwork Protocol ). P'rimeren je tudi za pov£»zavo računalnikov z r a z 1 i č n i m i r, : :i e r a c i j s i m i s i s t e m i .
Pravzaprav UNIX žele z novimi verzijami omogoča izvedbe kompleksnih mrež s skupno uporabo datotek in peri+erne strojne opreme. Funkcije in mehanizmi, ki omogočajo enotno vpeljavo različnih protokolov in podpor-o različnim medijem prenosa in ki so prej manjkali v operacijskem sistemu, so zdaj končno na voljo. Novosti "streams" in "sockejts" (5) sta dodaten mehanizem poleg že obstoječih krmilnih programov ( "device drivers" ). ATS'.T je npr. razvil poenoten vmesnik transportnega nivoja ( Transport Level Interface ), ki nam omogoča, da lahko pižemo aplikacije, ki so neodvisne od mreže. Zato se tudi nova verzija uucp-ja lahka izvaja neodvisno od mreže, ki se nahaja v spodnjih nivojih. S tem pa je uucp postal le že ena od možnosti uporabe mreže na najvižjem, to je api i kacijskem nivoju. Družbo mu delata npr. že dva protokola, ki sta od spodnjih nivojev neodvisna in ki omogočata preprost in hiter dostop oziroma souporabo datotek - RFS C Remote File Sharing, AT&T ) in NFS ( Network File System, Sun Microsystems ).
Povrnimo se že k mreži EUnet, saj se žiri tudi k nam. Letos osnovaiin Združenje uporabnikov UNIX tehničnih aplikacij ( ZUUTA ) pripravlja javno uucp mrežo, l.:i bo zaživela jeseni. Uporabniki mreže si bodo poleg elektronske požte	lahko	medsebojno	izmenjevali
najrazličnejše novice, iortnaci je, izkužnje in programe. Možna bo povez^iva s svetom, na domačem terenu pa bodo imele možnost tudi delovne organi zaci je, ki so zainteresirane za napredek na tem področju, saj bodo imele možnost stalnega kontakta z uporabniki.
Viri :
1.	XENIX Communications Guide, 1984 Intel
2.	A. Weiskopr: Networking Products Break New Sround, Uni k/World, 1/1987
3.	Quaterman i'. Hoskins: Notable Computer Networks, Commu.n i cat i ons o-f the ACM, Vol.29, Num. 10, 1986
4.	Houdler: EUnet, EUU6 Newsletter, Vol.7, Num.2, 19B7
■J. Cof-field Sheperd: Tutorial guide to Unii< sockets for network communications, Networks, Vol. 10, No. 1, 1987
6.	UNIX System V - Release 2.0 Administrator Buide, AT?-.T, 1984
7.	Zimmerman, H.:OSI Reference Model, IEEE Trans. Commun. Vol. C0M-2a, April 1980
B. The ROSE Consortium: ROSE, EUNET and the Migration to OSI, EUUS Newsletter, Autumn '86
7. Sklepne misli
Videli smo, da uucp zagotavlja možnost povezave
v mrežo za računalnike UNIXove bo počasi izgubil svoj pomen, ISG/OSI referenčnemu modelu projekti, ki imajo za cilj ne UNIX mreže ISO in CCITT ta način tudi z a gotovi ti r a č u n a 1 n i k i , k i i m a j □ sisteme. Tak razvojni, f:)r>: ki ga p o d p i T' a E S P RIT
družine. Kot tak riiaj ne ustreza y teku so i j ne sam o pi lag o d i t i st andar■Jfim, teiiiveč na kvalitetno povezavo z d r u g a č n e o p e r a c i j s k e jekt je npr. ROSE (8), 1 n f o !'■■ m a t i o n E! :< c: h a n cj e
INFORMACIJA iN INFORMACIJSKI SISTEM*
INFORMATICA 4/1988
UDK 519.724
Anton P. Železnikar Iskra Delta
v eseju se obravnavajo tale naslovna vprašanja: odpiranje nove. (informacijske) konverzacije, filbzofsko in biološko ozadje informacije (informacija kot oblika, vsebina, proces in pojav; ontološko razumevanje informacije; koncept informacije kot nastajajoče in razvijajoče pojavnosti), informacija kot princip in kot sistem- v živem, umetni informacijski sistemi (ai'Lificialno kot folozofija, znanost in tehnologija; načrt in izvedba informacijskega sistema; informacijska povezava človek-okolje in ciovek-stroj), primeri živih in umetnih informacijskih sistemov (možnosti formalizacije in konstruktivnega obravnavanja informacijskih sistemov in pojavov).
INFORMATION AND INFORMATION SYSTEM. This essay deals %-iith the following topics: opening of a new ( informational ) conversation, philotsophical arid biological background of information (information as form, contents, process, .find phenomenon ; ontologica'! understanding of information: a concept of information viewed through phenomenology of arising and developing), information as a principle and as a living system, artificial information systems (the artificial as philosophy, science, and technology ; planning and designing of an information system; i n foi-ma t i ona 1 conectedness in. the ma n - en v i ronmen t~ and man-machine interaction), some examples of living and artificial information systems .(possi hi li ties of formai ization and constructive approaches in the domain of i n f orma t i on systems and informational phenomena).
0. Uvod
Zival seveda ve. Vendar zanesljivo ne ve, da ve.
Teilhard de Chardin
a	(glej pojasnilo na koncu spisa)
Anton P. Železnikar
Priročna knjižica [IJ, ki jo je izdalo podjetje IBM, ima naslov: "Use Je informacija." Fred I. Dretske [2] začenja svojo bolj filozofsko kot tehnološko in matematično št\idijo o znanju in informacijskem pretoku z doma], a-gnostično premiso:' "V začetliu je bila informacija. Beseda je prišla kasneje. Ta prehod se je izvršil z razvojem organi žmovz njihovo sposobnostjo izbiranja Informacije za preživljanje in za nadaljevanje njihove vj-ste. " Hkrati pa ugotavlja, da "... se nahaja semantika izven obzorja matemat iòne teorije ih formaci je" in da
♦ Ta spis je bil predstavljen na Posvetovanju o informacijskih sistemih na Tehniški fakulteti v Mariboru, dne 2.6. 1988 in objavljen v zborniku posvetovanja.
"... nam komunikacijska teorija ne pove, kaj Je in formaci ja, Ta teorija zanemarja vprašanja, ki zadevajo ' vsehino signalov in kakšno informacijo nosijo, da lahko opisuje, koliko informacije nosijo. " Morda je pomembno, da v okviru Dretskejeve študije omenimo tudi njegovo stališče o znanju: "Kaj je znanje? Tradicionalni odgovor pravi, da je znanje oblika zagovurjanega prepričanja. ... Prepričanja so lahko lažna in resnici se lahko ne verjame. ... Ne pomaga tudi izgovor, kot se to večkrat dogaja, da je znanje odvi sno od ustreznega zagovarjanja, če se pri tem ne pove, kaj Je v svojem bistvu t.i. ustrezno zagovarjanje."
Zakaj je potrebno vsa ta vprašanja navajati in kako jih je mogoče vmeščati v diskurz vpraševanja o informaciji in o informacijskem sistemu? Človeško bitje, ki velikokrat zagovarja svoja razmišljanja in vedenje s t.i. zdravim razumom, ostaja pogosto izven domene vpraševanja o informaciji. Ali se kdaj samokritično sprašujete nekako takole: "Kaj je t.i. zdravi razum kot informacija ra zm iš1 j a j oče gà in v kulturno okolje prilagojenega bitja? Kako bitje v sebi .kopiči, prečiščuje, izbira in preoblikuje infoi-macijo kot, in formaci jsko bitje, kaj z njo počenja in kako pri tem uporablja drugo ■svojo informacijo?" Tu nastane vprašanje, s katero drugo informacijo in s katerim bitjevskim
informacijskim sistemom. Torej, ali ima bitje svojo informacijo in informacijo, s katero to svojo informacijo oblikuje, jo sprejema in sporoCa? Ce se o tem pogovarjamo s svojim zdravim razumom, nam ta odgovarja, da vsekakor imamo svojo informacijo in svoj sistem informacijske obravnave. Do naslednjega, bistvenega spoznanja je lahko le kratek korak: v bitju informacija nastaja kot življenska procesnost, kot regularna in neregularna (nova, mutue ijska) pojavnost živega organizma. Tako sproti in nenehno oblikujemo informacijo ne le za tiste funkcije, ki ohranjajo življenje, temveò tudi za drugo svojo aktivnost, za vedenje, ra^zmiš-Ijanje, sporočanje in informacijsko sprejemanje .
Kot informatiki, kot akterji in agenti na podroćju računalniških in informacijskih dejavnosti, kot socialna bitja informacijske epohe pa se nenehno soočamo s vpraševanjem, kaj je informacija in kakšen je njen sistem. Poudarjam, kakšen je njen sistemi Kakšne principe informacije je mogoöe konstruirati, če ti ne čakajo, da bi lahko bili odkriti, temveč so kot principi, kot principska informacija lahko le informacijsko konstruirani? Ce tega vpraševanja ne izvajamo, smo dejansko še predljudje, živimo bolj nezavedno in naključno vrženo v okolje in v sami sebe in ne spoznavamo dovolj inteligentno t.i. lastne in socialne, modsubjektne informacijske pojavnosti. Informacija prav gotovo ni nekaj enostavnega, statičnega, neprepletenega, necirkul i i-ajoči'ga, j(ì v stalnem nastajanju, kot da že tudi sama sebe oblikuje kot sistem. Je kot gost in gostitelj, kot objekt in subjekt v živih, nastajajočih in nenehno se spreminjajočih informacijskih agregatih. Informacijska doba prinaša novo razumevanje informacije, ki je nova razvojna stopnja človeka kot visoko strukturiranega in organiziranega bitja. "Eden od namenov tega eseja Je odpiranje nove konverzacije, s katero bi bilo mogoče artikulirati moč: človekovega informacijskega uma," je bilo rečeno v [3]. Poudarek te izjave je v tem, da se končno informaciji kot fenomenologiji živega prizna tista njena narava, ki je že tisočletja spoznavno zasidrana v zdravem razumu človeka in ki sega iz pradavnine, starogrške filozofije, gnosticizma, evropskega racionalizma do novoveške filozofije in se danes kot še nikoli doslej pojavlja kot nov, zavesten in smiseln imperativ preživetja.
1. Filozofsko in bioloSko ozadje informacije
1.0. Informacija kot. oblika, vsebina, proces in pojav
Za živo bitje ima informacija pomen predvsem kot pojavnost v njegovem organizmu. Ta pojavnost, ki jo imenujemo informacijska, ima svojo oblikovnost (genetičnost, strukturnost), vse-binskost (odnosnost, organ i zi ranost) oziroma kratkomalo procesnost. To pa sploh ne pomeni, da ne obstaja tudi neživa informacija, ki jo lahko opažamo v naravi, v kozmičnih pojavih in v človekovih tehnoloških in tudi že v bioloških artefaktih. V področje umetne informacije uvrščamo tudi računalniško bazo podatkov, računalniški program in računalniško arhitekturo, ki omogočajo fleksibilno (programirano)
podatkovno transformacijo. S tem smo povedali le to, da je informacija informacijsko aktivna in pasivna pojavnost vesoljnega (naravnega), življenskega (biološkega) in artificialnega (konstruktibilnega, umetnega).
Informacijska oblika in informacijski proces sta vselej tudi vsebinska (pomensko spremenljiva in odvisna od razumevanja, ki ga bitje trenutno premore). Informacijski pojav je kot oblikovna in procesna entiteta (avtonomnost, združenost, zaokroženost, vase-zaključenost ) nosilec: takšne ali drugačne vsebinskosti (interpret i vnos t i , kodiranosti ) . S pojavom oblike in procesa se odpira neizčrpen in neponovljiv videz informacije. Oblika in proces lahko nosita (povzročata, vznemirjata) pojavno raznovrstna sporočila in začenjata nove, nastajajoče informacijske pojave in i n f o r mac i j sito oblikovanje (kreiranje). Informacija pomeni individualnost tako po obliki in/ali procesu. Senzorska informacija bitja se na svojih poteh do korteksov individualno (svojsko, nepredvidljivo, metafizično) preoblikuje v odvisnosti od obstoječe, totalne bitjevske informacije. Pojav, ki kot opaženo nekaj sporoča, se različno informacijsko interpretira v organizmu bitja in sproža različne informacijske akcije (reakcije, odločitve, voljo za vedenje, zavest). Informacija prinaša tako vselej še potencialno, neo pa žon o, neraziskano komponento in tako učinkuje na informacijsko okolje (domeno, carstvo), v katero vstopa.
Sodobna tehnologija in spoznanja v raziskavah procesnost! žive celice prinašajo zavest o i nformaci jsltem kozmosu, ki bi bil v tehnološkem in živem potencialno možen kot svojski (informacijski) sistem (nevralni, genetski in masivno-paralelni računalniki). Živa celica ima kompleksen, individualen, avtonomen informacijski sistem, s svojo logiko (dednostjo, mutacij-skostjo, fleksibilnostjo) in tako svojsko procesnost, ki je značilna za življenski cikel celice, za njeno preživljanje, obrambo (imunost), prehranjevanje (metabolizem), replikaci-jo (razmnoževanje) oziroma samoprodukci j o (avtopoiesis). Na primeru žive celice je mogoče spoznavati, kako so v biokemijsko informacijsko st.rukturo (molekularno inteligenco) vmeSčeni oziroma vgnezdeni drugi informacijski agregati, in sicer od obi ikovno-pomni 1 ni h (dednostna struktura DNA, RNA) do biološko-procesnih. V računalniških informacijskih sistemih imamo prav tako značilne pojave vmeščanja ali vgnez-devanja ene informacije (podatkovne strukture) v drugo. Tako je npr. izdelani uporabniški program najprej vmeščen v okolje jezikovnega prevajalnika, prevajalnik v okolje operacijskega sistema in operacijski sistem v okolje računalniške arhitekture. Vsi ti elementi račiuialnika so razumevani kot specifična informacija, in sicer za trenutno uporabo, prevajanje, operativnost in strojnost. Pri podatkovnih strukturah kot definicijsko statičnih informacijskih agregatih je princip vgnezdova-nja lahko rekurzivna lastnost podatkovne strukture .
1.1. Ontološko razumevanje informacije
Ontološko ali bitjevsko razumevanje informacije je v svojem filozofskem ozadju lahko fenomenološko, eksistencialno in funkci ona 1 i s t ično. Razumevanje nastajanja informacije je pojmovni paralelizem k nastajanju živega in je neposredv no s tem razumevanjem povezano. Ontološko razumevanje informacije je preprosto vpraSanje: "Kaj Je informacija." Iz tega zaCetnega vprašanja se lahko razvija proces vpraSevanja, ki se razširja v kajstvo biti informacije oziroma v vpraševanje o informacijski biti.
Človek zmore gledati zavestno v svojo ali tujo informacijo in jo zmore tudi zavestno opazovati , raziskovati, spoznavati in razumevati. Tu je doloCena informacija objekt Človekovega miöljcnjn, nnmo miflljonjo pn infornmci j h k i subjekt, ki zmore opazovati svoje informacijske objekte. V tej svoji informacijski funkciji lahko mišljenje nastajalno (generativno), spontano (nepredvidljivo), cirkularno (reku-rentno) in paralelno (prostorsko in Časovno hkratno) transformira (manipulira in preoblikuje) informacijske objekte, ki so oblike in procesi kognitivne informacije. Tako nastaja ali se poraja vzvratno prepletena informacijska interakcija med informacijskimi subjekti (operatorji) in informacijski objekti (operandi), ko se hkrati izmenjujejo njihove objektivne vloge s subjektivnimi in obratno. Mišljenje se očitno manifestira (aktivno in pasivno informira) kot operacija in operand mišljenja. V človekovih korteksih je tako mogoče zavestno (introspektivno) opazovati subjektivno-objek-tivno procesnost informacijskih pojavov. Prav od tu, iz tega primera je tudi mogoče spoznavati spontano, nepredvidljivo, generativno, cirkularno, paralelno, posledično in vsakršnje nastajanje informacije v živem.
Ontološko razumevanje informacije je tedaj utemeljeno s prepričanjem, da je informacija tudi pojavnost živega in da zaobseže vprašanje o biti informacije prav za prav vsa druga vprašanja, ki se nanaSajo na bit (nem. das Sein, angl. Being), na njeno eksistenco in njeno nastajanje..V okviru informacijske spoznavne zavesti je vse pojavljajoče in nastajajoče informacijsko fenomenološko, pri čemer pii se lahko v okviru filozofije informacije postavlja tudi vpraSanje o informaciji po sebi. Seveda pa se z informacijskega vidika informacija kaže tudi v vseh kulturnih oblikah, npr. kot filozofija, znanost, umetnost, zavest, hotenje, prepričanje, umovanje, čustvovanje, sklepanje itd. Z informacijsko ontološkega vidika obstajajo le značilne informacijske oblike in značilni informacijski procesi, med katere se uvrščajo tudi vse kultijrne, ideološke in religiozne pojavne oblike. Bitje se ob vseh teh informacijskih oblikah (kulturnih, socialnih, okoliških, kozmičnih) lahko identificira le, s svojo celostno informacijo, ki jo je vobče mogoče označiti kot bitjevsko metafiziko. Taksno, kibernetsko pojmovanje informacije obstaja v bistvu (nezavedno) že v začetkih grSke filozofije, v novejši dobi pa se informacija kompleksno k iberne t i čno , seirilotsko in sistemsko pojmuje že vsaj od.nastanka Wienerje-ve kibernetike.
1.2. Koncept informacije kot nastajajoče in razvijajoče pojavnosti
Pri pojmovanju informacije nam gre torej za razumevanje prav tistega, kar sproti spontano, nepredvidljivo in seveda tudi pričakovano nastaja ali je šele v razvoju v živem (organi-zmičnem, bitjevskem), vesoljnem (kozmičnem) in artificialnem (umetnem, konstruktibiInem) in se pojavlja kot informacijsko. Informacijsko v živem je sedanjost nastajajočega iz preteklega (zgodovinskega) in iz videnja prihodnjega (prihajajočega). Bitjevska metafizika razpolaga s svojim modelom sveta (modelno informacijo) in s svojimi mehanizmi za upoštevanje tega modela (idelogije) pri nastajanju bitjevske informacije. Tudi nizmiSi jim je o informaciji je lahko le metafizično, torej bitjevsko informacijsko, Hi'vcila , il i niun i čno [i f('p I l.ono in vzvratno vna-p r- (■■! j S n j e . Razmišljanje je lahko tudi bolj statično, vprežono v znanstveno disciplinar-nost, v ideološko omejenost, tako da je npr. deduktivno, aksiomatično, linearno sklepajoče, konsistentno, neproti slovno in v okviru svoje omejenosti tudi produktivno. Informacijska doba prinaša prav izziv informacijskega, svobodno, pluralno nastajajočega.
Razumevanje informaciske nastajalnosti pa prinaša nove možnosti prav tam, kjer so se doslej pojavljale nepremostljive ovire: v znastvenem, raziskovalnem in tehnološkem. Inteligenca kot informacijska lastnost živega je zaenkrat pojavnostno le nejasen ideal (cilj), ki se mu tehnološko in znanstveno poskušata približevati skozi di sciplinarnosti umetno-inteligenčnega, spoznavno-teorijskega, nevroznanstvenega, psihološkega in filozofskega. Človeka vznemirjajo danes raziskave inteligenčnega, žive inteligence, ki je kognitivno in spontano nastajajoča in se pojavlja kot bitjevska .vrženost (nem. die Genorfenke i t , angl. throwness) v življensko problematične (usodne, nepredvidljive) položaje. Seveda je v tem pomenu informacijskega, zlasti nastajalnega prej ko slej potrebno razumevati (predvidevati) tudi drugačne (nove) možnosti abstrakcije, simbolične' formalizacije in njene pomenskosti, ki naj bi presegale trdna, znastveno statična in s tem varna in produktivna (?l ) zatočiSča rac: i onal i s t ičtie t rad i c i j o .
2. Informacija kot princip in kot sistem
2.0, Informacijski principi
Filozofske principe informacije je mogoče .opredeliti intuitivno in jih potem (glej [4] in v tem spisu odstavek 4.2) tudi simbolično formalizirati. Poskusimo takole:
(1)	Informacija je cirkularen in spontan pojav informacije, pri čemer sta tudi njena cirkular-nost in spontanost informacija. Dva informacijska pojava (informacije, informiranja) sta vobče (zaradi njune spontanosti) informacijsko različna.	□
(2)	Informacija nastaJa z informiranjem, in sicer s pròti inform i ran Jem in z vmeščanj em
nastale informacije v obstojeCo informacijo. Tudi nastajanje informacije je informacija. □
(3)	Nastala informacija je pro ti i nf ormaci Ja (nova, še ne vniešCena informacija).	G
(4)	Informiranje informacije je prihajanje informacije v obstajanje (v eksistenco), njeno spreminjanje i ti izginjanje; to pii jo prav 1'niio-menoJogiJa informacije.	n
(5)	Z vmeéóanjem nastale informacije oziroma protiinformacije v obstoječo informacijo se nastala informacija oiiranja, ker se sicer izgublja oziroma se pojavlja kot t.i. informacijski šum.	D
(6)	Cirkularnost in spontanost (nastajalnost) informacije sta transparentni za celotno domeno informacije, tj. za informiranje, proti informacijo, protiinformiranje (informacijsko diferenciranje), vmeščanje (informacijsko integriranje) itd. Cirkularnost informacije je njena rekurentnost, paralelnost, zaporednost (posle-diònost) itd.	□
(7)	Paraleinost/serialnost informacije je topološko, prostorsko, časovno generativna (nastajalna), spontana cirkularnost.	O
(8)	Struktura informacije so informacijsko sestavljene informacijske oblike in informacijski procesi. Organizacija informacije so informacijski odnosi (funkcionalne informacijske povezave) med njenimi oblikami in procesi. Struktura in organizacija informacije sta informacija in oblikujeta t. i. integracijo informacije (integrirano informacijo).	□
oziioma od enostavni)! do zapletenih informacijskih funkcij. Informacija se kot oblika in proces (kot biokemično pogojno stabilna in spremenljiva struktura in organizacija) pojavlja (razumeva) kot sistem v živem sistemu in tudi kot informacija o samem živem sistemu. RaKUmevanje informacije kot svobodno dinamičnega procesnega sistema je tako usklajeno z današnjim razvojem zdf!iv«ga razumu, je kiberne-l.ično nast-ajiilno in	iijniiin z raznovrstnimi
znanstvenimi, izkustvenimi in filozofskimi spoznanji in tako ostaja na poti aktualnega spoznavanja.
3. Umetni informacijski sistemi
3.0. Artificialno kot filozofija, znanost in tehnologi Ja
Prihodnji človekovi stroji bodo Se vedno njegovi artefaltti, čeprav bodo imeli strukturo in organizacijo t.i. nevralnih vezij in s tem svojo lastno genetičnost, samoorganiziranost in samorazmnoževalnost. To naj bi bili stroji učenja zlasti na področju njihove uporabe. Vendar bo tudi tem strojem potrebna neka rudimentarna, začetna, izhodiščna struktura in organizacija in v tem segmentu njihovega konstruiranja in nastajanja bo potrebna informacijska filozofija, fundamentalna (strukturna) znanost in nova (biološku) tehnologija. Ti novi artefakti bodo s človekom lahko povezani tudi signalno, preskakujoč človekove senzorje (oko, uho, tip).
(9)	Informacijski stroj ima strukturo in organizacijo informacije (npr. informacijo v okviru informacije in stroj v okviru stroja).	□
(10)	Informacijski program informira spontano in cirkularno, nastaja s proti informiranjem in informacijskim vmeščanjem, se vede kot informacija v okviru informacijskega stroja. Informacijski program more modificirati samega sebe in druge programe in more biti. modificiran z drugimi informacijskimi programi med svojim izvajanjem in izvajanjem drugih programov. □
(11)	Me ta fi z ika je označevalec - za celostno (totalno) informacijo bitja. Metafizične komponente so npr. model sveta, čustvena, senzorska in motorna informacija bitja. Bitje je mogoče razumevati kot avtopoietski informacijski sistem.	□
2.1. Informacija kot sistem v živem sistemu
Po tem, kar je bilo doslej zapisanega, se informacija pojavlja kot svojski (spontan, cirkularen, nastajalni, prepleteni) sistem. Kaj je v informacijskem sistemu kot pojmovanju sistema bistveno različno od samega pojma informacije? V okviru živega organizma oziroma bitja je informacija sistem procesov, ki se pojavlja v strukturi in organizaciji živega. Vsi ti procesi prinašajo, preoblikujejo, rojevajo in namenjajo informacijo in delujejo sami kot informiranje (procesiranje), od najnižjih molekularnih, celičnih do korteksnih ravnin
3.1. Načrt in izvedba informacijskega sistema danes in Jutri
Današnji informacijski sistem je opredeljen kot sistem, ki Je bil razvit za oblikovanje (kreiranje), zbiranje, shranjevanje, procesiranje, razdeljevanje in interpretacijo informacijskih množic [6]. Današnji informacijski sistem je tako potrebno razviti pred njegovo uporabo. Ta razvoj vključuje analizo potreb, načrtovanje sistema in njegovo realizacijo.
Človeška bitja kreirajo in interpretirajo informacijske mno-žice informacijskega sistema. Informacijski sistem nima vrednosti sam po sebi, če ne olajšuje, pospešuje in izboljšuje človekove aktivnosti. Današnji informacijski sistem tako še nima vgrajene tiste funkcije učenja, ki bi avtomatično zagotavljala njegovo prilagovanje potrebam uporabnika. Ce to sicer velja za človeški faktor sistema pa to prav gotovo ni res pri strojnem delu informacijskega sistema. To pa zahteva nenehno kreativno poseganje (analizo, načrtovanje, realiziranje) človeka v sistem. Jutrišnji informacijski sistem naj bi bil predvsem strojno informacijski tako, da bi interaktivno razvijal svojo strukturo in organizacijo in da ne bi zahteval bistvenega poseganja človeka. To svojo funkcijo samoorganizacije naj bi stroj v sistemu opravljal z rudimentarnimi funkcijami učenja, svoje samostrukturiranje pa naj bi dosegal v okviru tehnologije nevralnih vezij.
3.2 Informacijski povezava človek-okolje in človek-stroj
Informacijska povezava človek-okolje se kaže kot povezava med človekovo metafiziko ter senzorsko in motorno informacijo.. Ti trije informacijski pojavi (metafizika, senzorska in motorna informacija) se prepletajo in vzajemno učinkujejo na informacijsko nastajalnost. Simbolično (glej kasneje) lahko to prpcesnost nakaZemo s tremi definicijami:
nastajanje metafizike |ji
"Df
nastajanje senzorske informacije cr
"Df
nastajanje /notorne informacije X
~Dt
o-, A	(J.
<T, A , l-i 1=
A , u, o- 1=
Ji, (T in A oblikujejo človekov paralelni inferra ac i j s k i s i s t e ra . Stroj (računalnik) je le posebno okolje, ki podpira človekov informacijski sistem in opravlja zanj specifične podatkovne naloge hitreje, zanesljiveje in algorit-mično "objektivno". Povezava človek-stroj je danes vizualna, akustična in taktilna. S pojavom t.i. nevralnih in mas ivno-paraleInih računalnikov pa bo lahko postala tudi signalna (nevrofiziološka).
pojavljajo v živi celici, kjer oblikujejo celični citoskeletni (raikrotubulni) informacijski sistem [5], od katerega sta odvisni samoorganizacija in samoprodukoija cel ice. 'Populacije živih celic sestavljajo organizme, ti pa višje oblike organov vse do človekovih korteksov. V korteksih sta tako struktura kot procesnost informacijski. Nevroni so osrednji procesorji, sinapse pa kompleksni povezovalni procesorji med nevroni. Nevronska mreža se oblikuje po dednostnih zasnovah in tudi pod informacijskimi "pritiski", ki so posledica dogajanja v okolju in interakcije bitja z okoljem. Raznovrstnost je tu očitna: dva nevrona nista enaka, pa tudi ne dve sinapsi in živa nevronska mreža nima dvojnika v svetu živega. Ob vsem tem se trenutni informacijski procesi v nevronski mreži prava značilnost oziroma neponovljiva individualnost dane- mreže. V carstvu žive informacije, od celične, organizmične do korteksne, velja ekskluzivni princip pluralnosti, enkratnosti informacijskili individuov in njihove neponovljivosti. Vsak živi oziroma Zivljenski informacijski sistem je unikat po oblikovnosti (biološki strukturiranosti) in po procesnosti (biološki organiziranosti). Princip informacijske pluralnosti velja tedaj že za molekule življenja (prostorske vijačnice), celice (mikrotubul-ske vrtiljalce) in organizmične podsisteme (nevronske populacije).
4.2. Simbolna formalizacija in formaci Jskih sistemov
4. Primeri umetnih, živih in abstraktnih (simbolnih) informacijskih sistemov
4.0. Umetni informacijski sistemi
Umetni informacijski sistemi so tehnološki sistemi upravljanja ali pa so z živimi bitji povezani sistemi, v katerih se interaktivno upošteva tudi živa, bitjevska informacija. Cisti tehnološki sistemi seveda informacijsko razumljeno ne morejo biti pravi informacijski sistemi, saj je pri njih le težko mogoče govoriti o tistem nastajanju informacije, ki je za informacijo glavna značilnost. Ti sistemi so predvsem algoritmični , rac ionalistični in imajo kvečjemu pogojno predvidljivost, ki pa nikakor ne more nadomeščati informacijske nepredvidljivosti. Takim sistemom lahko rečemo podatkovni sistemi. Mešani sistemi so živi informacijski sistemi s tehnološko podporo in organizacijo; informacijski so prav zaradi žive komponente, zaradi svojih nepredvidljivih akterjev, ki v te sisteme funkcionalno posegajo, jih gradijo in modificirajo, prilagoditveno kreirajo v njihovih življenskih ciklih. Umetni informacijski sistemi so tako artefakti človeka z njegovo informacijsko udeležbo v njihovi funkciji.
4.1. Zivi informacijski sistemi
Formalizacija informacijskih procesov in sistemov, ki so sestavljeni iz takih procesov, je mogoča z vpeljavo posebne, informacijsko specifične simbolike, in sicer v obliki t.i. informacijske logike [4]. Pokažimo samo formalizacijo z uporabo t.i. splošnih operatorjev informiranja .
(1) Informacija a informira
'Df
a	v H a
informacija a Je informirana
"Df
a v a =(
(2) Operatorja informiranja f= in =j sta splošna operatorja informiranja in ju je mogoče parti-kularizirati (npr. v znane logične, aritmetične, modalne operatorje) in univerzalizirati na poljubno kompleksen način. Tako imamo npr. tele pomene :


pomeni	a informira pravi Ino;
pomeni	ß se verjame ;
« pomeni	verjame se, da se a verjame ;
pomeni	verjame se, da a verjame itd.
Operatorja f= .in =| sta lahko unarna, binarna ali multipleksna, infiksna, prefiksna ali postfiks-na in .predstavljata neomejeni, nastajalni, genei'ativni množici operatorjev (z njuno parti-kularizacijo in uni verzal i zaci jo).	■
Zivi informacijski, sistemi obstajajo na različnih ravneh ž ivobitjevskega. Določene informacijske mehanizme je mogoče opaziti že v t. i. molekulah življenja, kot so DNA, RNA in seveda beljakovine. Te molekule se med drugim značilno
68
(3) Informacija a ne informira
'Df
at* A	c<
Informacija a ni informirana
'Df
t* a A aH
(4) S formalnim orodjem, ki smo ga opredelili, je mogoče interpretirati simbolično metaforo na samem zaCetku spisa. Ta metafora izhaja iz izpeljave tipa modus ponens takole:
(1= a) h
a= a ^ q J) (-
( (N a H (7)	( q 1= «) )
( ( C/ a j= a) A ( a t= q) )
( ( a, a t= C. a. « ) "I )
h h
Seveda bi bilo mogofe izraziti tudi bistveno drugačne in bolj zapletene scenarije vpraševanja.	■


To pa je prav de Chardinov izrek, kjer je ci metafizika (vedenje) živali, oznafuje 'zanesljivo ne vedeti' in l=j, 'seveda vedeti'.	■
(5)	Informacijske operatorje je mogoče partiku-larizirati in univerzalizirati poljubno na že znan pomen ali na nov, tudi informacijski pomen. Smiselno je mogoče vpeljati 16 specifično informacijskih operatorjev in jih seveda partikular!zirati po potrebi, nekatere med njimi pa tudi univerzalizirati. Tako je mogoče uporabljati splošne ((=, , =j, t^),), paralelno splošne (Ih	4), ciklične ( [-. V, H, ) in paralelno ciklične (|h, it/,	HI) informacijske operatorje. Na osnovi teh operatorjev je mogoče konstruirati generativni aksiomatični sistem oziroma t.i. informacijsko logiko, ki ponazarja predloženi koncept informacije kot nastajajoče fenomenologije. ■
(6)	Nazadnje si oglejmo Se primer opisa procesa, ki ga v vsakdanjem pomenu imenujemo vpraševanje. Vpraševanje je interaktiven informacijski proces (diskuiz), v katerem se pojavljajo zadevna informacija a, nn katero bo vpraševanje naslovljeno, pojavljajoče (začetno in razvijajoče) vprašanje q, in nastajajoči odgovor a. Posamezne faze procesa vpraševanja so ciklične (operator h), v ciklih pa se pojavlja splošno informiranje (operator |= ) . Pojavitev vprašanja q (njegovo prihajanje v obzorje informacije a) je označena z operatorjem začetka (J), končanje procesa vpraševanja pa z operatorjem konca (T). Primer vpraševanja je tako tale:
5. SLOVSTVO
[1]	VOLLMER, R., Alles ist Information, IBM
Enzyklopedie der Informationsverarbeitung, IBM Deutschland, 1984.
[2]	DRETSKE, F. I., Knowledge and the Flow of
Information, Basil Blackwell Publisher, Oxford, 1981.
[3]	ZELEZNIKAR, A.P., Principles of Information, Informatica 11 (1987), No. 3, 9-17 .
[4]	ZELEZNIKAR, A. P., Information qua Information, Zbornik radova Mipro '88, New Generation Computers, 4/15-30.
[5]	HAMEROFF, S.R., Ultimate Computing: Bio-
molekular Consciousness and Nanotechnolo-gy, Elsevier Science Publishers (1987).
[6]	LUNDEBERG, M., G. Goldkuhl, A. Nilsson,
Information System Development, Prentice-Hall Inc., Englewood Cliffs, NJ (1981).
(V Ljubljiini, 22. mnja 1988)
FORUM INFORMATIONIS
FORUM TNI'OUMATTONr.S V ČASOPISU I N f-'OlìMATi CA
V tej Številki Informatice objavljamo v rubriki Forum inforraationis kratko poročilo o zasedanju Forum i riforma t ion I k ,	ki jo; obravnav.! lo
problematiko	računalniške	industrije,
nadaljujemo z objavo odgovorov na odprto pismo iz prejšnje številke in naposled prinašamo Se primer zares sporne domaČe strokovne recenzije, ki pomeni bržkone afero par excellence na področju strokovnega in mentorskega dela Ril.
KRATKO POROČILO Z DRUGEGA ZASEDANJA FORUM INFORMATIONIS
Drugi sestanek FI za okroglo, mizo, dne 29. junija 1988, v sejni dvorani SOZD Iskra, v, Ljubljani, Trg revolucije 3/XIV, je bil namenjen problematiki računalniške industrije v Sloveniji. Na ta sestanek je bilo vabljenih 50 strokovnjakov, novinarjev, zasebnikov itd. s področja računalništva in informatilie (skupaj 28 podjetij in 3 zasebniki). Udeležba je bila s 17 udeleženci dokaj pičla, in sicer: Jože Buh (I Delta), Rado Faleskini (SOZD Iskra), Matija Hudovernik (Radio Ljubljana), Stanislav KleSnik (Intertrade), Marko Kovacevic (Mikra), Vinko Kurent (Mikroada, Maribor), MiloS MarinCek (FAGG), Marijan MiletiC (IJS); Saša PreSern (I Delta in IJS), Mija Repovz (Delo), Slavko Ro2ic (I Delta), Brane Semolic (Gorenje, SOZD), Peter Stanovnik (Institut za ekonomske raziskave), Maks Svensek (Birostroj, Maribor), Ivan Santi (I Delta), Jure Spiler (inovator v svobodnem poklicu) in Anton Zeleznikar (Časopis Informatica).
Vabilo za to zasedanje FI je vsebovalo tale izhodiščna vprašanja.: Kaj, kako in koliko proizvajati? Zakaj je lastna računalniška industrija smiselna (potrebna, perspektivna)? Kje so možnosti proizvodnje za izvoz? Kako je z zaščito industrijske in intelektualne lastnine? Kam plove slovenska računalniška industrija? Ali se je že'približal trenutek streznitve, ki bi lahko pomenil delitev dela, organizacijo velikih in malih, novo pobudo za zagon industrijske miselnosti?
Oglejmo si na kratko povzetek najmarkatnejših misli, ki so bile izrečene na forumu.
Slovenija nima posebnih političnih in gospodarskih resorjev za področje Ril. Kako naj se v tej praznini razvija računalniška industrija, ki je prepuščena le svoji, večidel nestrokovni spontanosti? Slovenija npr. nima niti ■ ministrstva, niti. instituta, niti strokovnega odbora pri GZS za Ril. Celo ponekod v manj razvitih delih Jugoslavije nekaj teh služb imajo. Organizacijska zavest se tako le s težavo oblikuje in ostaja prej ko slej samo koncept kratkoročnega preživljanja. Npr. naCrt plasmaja izdelkov domaČe	računalniške
industrije na Zapad praktično ne .obstaja, oblikuje pa se trenutno se vedno rèntabilni plasma na Vzhod. Pri tem je delež tujega znanja v domaČih izdelkih ne samo prenizek temveC tudi premalo inovativen, tako da je domaČe znanje v glavnem že nerelevantno.
V začetku 60-ih let se je v Sloveniji pojavilo podjetje Zuse K. G. ob zastopstvu IBM. Ti
poj,iVi začetkov slovenskega Ril so bili Hpri-iul j.IUI /. 1 cu/. i vi\ i flol.iiijiMu t-\ul i na univerzi. Ze leta 1971 je ■ bila Ljubljana gostitelj	svetovnega	kongresa	IFIP
(International Federation for Informastion Processing). 7.a to obdobje je bila značilna odločitev, da Slovenija stopa in stopi intenzivneje na pot Ril. Leta 1982 se je pojavil računalniški zakon, ki je nalagal dogovarjanje v razvoju in poslovanju Ril. Zakon je bil posledica občutenega razkoraka med domaČim in tujim znanjem, obrambna poteza preti (Ji-fid 1 i-.i jo<-i m ( r.m!-in,K: ii>v>.i 1 kam. Ta zakon je bil posledica samozadostne dobe. Medtem ko se je strategija transnacionalk bistveno spremenila, smo doma pozabljali na nova merila. Kritična zamuda je bila narejena v razdobju med 1970 1978. DomaČe neznanje je postajalo" Čedalje dražje in neracionalno.
Današnje znanstvene in tehnološke primerjave Ril upoštevajo skoraj izključno le stanje v ZDA in Japonski. T.i. revolucionarni tehnološki skoki se pojavljajo prav v teh dveh deželah zaradi njunega medsebojnega tekmovanja. Vendar ima računalniško industrijo ves razviti svet, tudi Indija. V procesu globalizacije, kjer postajajo veliki prizvajalci Se vecji, je opazno spajanje producentov in uporabnikov. V tej globalizacij i pa ostaja dovolj prostora tudi za majhne, saj jihvleCejo naprej veliki s sodelovanjem in z vsiljevanjem standardov. Tako se ustvarja nov prostor sožitja velikih in malih - podjetij, narodov in populacij. Slovenija (ali SFRJ) se lahko konkretno primerja s svojimi sosedi: Italijo, Avstrijo in Madžarsko. Računalniška industrija je tudi investicijsko intenzivna in dotpk tujega kapitala je nujen. Podjetje Honeywell je npr. že vložilo svoj kapital v proizvodne obrate pri nas. ,
Računalniška industrija v Sloveniji se sooCa s problemi financiranja nabave materiala oziroma s finančno nelikvidnostjo. V bistvu se ta industrija kreditira izključno s prednaročili uporabnikov. Hkrati se pojavljajo (tudi zaradi tega) dodatni problemi s principali (IBM, DEC). Slovenija v razreševanju te problematike Ze zaostaja, ker ne realizira nujnega prenosa tehnološkega znanja,	servisiranja	in
samostojnejšega razvoja programske opreme. Principali sicer kažejo veC razumevanja, vendar določenih pobud ne razumemo ali pa jih ne znamo izpeljati. Za neposrednejse sodelovanje z^ velikimi računalniškimi podjetji pa mora biti' izpolnjena vrsta zahtev, kot so (devizno) izšolani kadri, finančna sredstva za t.i. zagonske investicije, kvalitetna proizvodnja, resnost pri poslovnem dogovarjanju in zanesljivost dobavnih rokov. Skoraj ničesar od naštetega ne obvladamo! Verjetno bi bila najkrajša pot priucitve	neposrednej Se
sodelovanje s tujimi partnerji, zacensi z manj zahtevno proizvodnjo in potem naprej s čedalje zahtevnejšo.
V SFRJ obstaja in se Se vedno izdeluje strategija , tehnološkega razvoja, vendar Slovenija potrebuje tudi svojo, bistveno tehnološko razvojno strategijo, ki mora biti remek delo v primerjavi z jugoslovansko tehnološko globalizacijo. Izoliranost slovenske industrije proti razvitemu svetu je simptomaticna: nobeno podjetje ni včlanjeno v ECMA. Brez tega pa resno sodelovanje s partnerji iz evropskega prostora ni mogoCe.
Nujna je tista liberalizacija, ki jo poznajo npr. v Južni Koreji, Indiji, Finski in Avstriji; v teh državah je opazen tudi vpliv Vzhoda. In seveda, slovensko podjetje lahko v okviru večje gospodarske 1ibera1izacije uvaža le s svojimi deviznimi, si'edsbvi..
Svetovne posJovne in l.ehno J uftke normi-' niso nikakrsnji izumi kulturnih revolucionarjev. Tako je npr. avans posojilo in ne dohodek, inženir ustvarja novo vrednost; daje veC kot dobival Itd. Veliko tega, kar je svet napravil, je mogoče takoj pobrati. Gorenje ustvarja velik izvoz	z velikoserijsko (veCmi1 jonsko)
proizvodnjo. Strategija Elana kot uspešnega svetovnega ponudnika je, da je vselej samo drugi (ne prvi!). Računalniška industrija se lahko iz teh domaćih primerov marsikaj nauci. Tudi juznokorejske in celo mehiške industrijske serije dosegajo danes stotisoce kosov. V začetku so imeli skupne posle (joint venture) s transnacionalkami, danes pa imajo 2e lastne zaščitne znamke (trade marks). Proizvodnja, raziskave in razvoj se čedalje bolj internacionalizirajo. Pri zasciti domaČih trzisc so dopustne razlike v ceni le do 15%.
Primer Indije je zanimiv. Indija velja za najbolj demokratično družbo. Taksne druzbe imajo izrecne perspektive v svojem razvoju: znanstvenem, tehnološkem in gospodarskem. Indijska , demokratičnost je posledica izredno heterogene druzbe. Iz tega bi se bilo mogoče kaj naučiti tudi za jugoslovansko demokratično prakso, ki naj ne bi ostajala le izključna razvojna	prednost	Slovenije.	Večja
demokratizacija. naj bi omogočala tudi organiziranje male proizvodnje,	novih
institutov in podjetij, v katerih bi neizziveti potencial Jugoslovanov lahko prišel do polne veljave. Tako bi spontano nastajala tudi bolj aktualna kulturna izmenjava demokratične in civilne jugoslovanske misli.
Zasebniki in mala podjetja na področju Ril očitajo delovnim organizacijam (Delti) neodgovornost menaziranja. Ta neodgovornost na različnih upravljavskih ravneh v bistvu odbija sposobne kadre, ki odhajajo drugam in v inozemstvo. Omenjen je bil konkreten primer prof. Egona Zakrajska iz leta 1981; on je danes vodja razvoja podjetja Cromemco v Kaliforniji. Sploh je podobnih spornih kadrovskih primerov kar veliko. Prednost zasebnika je očitna: sam odloča, sam sklepa pogodbe, obraCa denar in ima opravka s sodiščem. Želel bi imeti podporo v kaki firmi za buticne proizvode. Zasebni inženirji računalništva imajo predvsem manj računalniških orodij in se zato težje znajdejo. Tudi za pridoVjivanje novecj.i znanja je v malih podjetjih poskrbljeno.
V Jugoslaviji se pojavlja tudi kriza t.i. zastopništva. Jugoslovanski trg stagnira, jug države se sooCa z nelikvidnostjo, prodaja je mogoča le Se za devizna sredstva pri velikih uporabnikih. Principali ne investirajo, tovarne se gradijo z lastnimi sredstvi le za domaci trg. Principali tudi ne dajejo znanja in ne ponujajo možnosti ižvoza za izdelke z njihovo tehnologijo! V Avstriji, Italiji in v deZelah Beneluksa država podpira uspešne proizvajalce prav pri investicijah (v Sloveniji sta npr. taka primera Intertrade in Delta). Vprašanje o potrebnosti računalniške industrije je najbrž preseženo. Računalniški proizvodi se kot deli in sistemi pojavljajo praktično povsod. Veliko-serijska računalniška proizvodnja pa je danes centralno regulirana oziroma omejena z devizno kvoto za material.
A. P. Zeleznikar
KRATEK KOMENTAR K DRUGEM ZASEDANJU FORUM INFORMATIONIS
To, kar je na drugem zasedanju FI prav gotovo zbodlo v oci, je bila odsotnost nekaterih glavnih akterjev na področju računalniške industrije v Sloveniji. Zaradi tega je bil diskurz o računalniški industriji morda za nekoga res tako abstrakten, kot je to občutila novinarka Mija Repovz v svojem prispevku "Medla abstraktnost", objavljenem v Delu, 7. julija 1988. Vendar moram glede na gornje poročilo o zasedanju FI prav zaradi tega postaviti vprašanje, za koga je bila lahko vsakdanja problematika računalniške (in druge) industrije, tako kot se je pokazala na FI, le medlo abstraktna. Očitno imamo opravka kar z znatnimi razlikami v razumevanju vsakdanje problematike industrije pri ljudeh v tovarnah in pri tistih, ki informirajo iz salonov nasih novinarskih his. Verjetno bi lahko novinar uglasil svoje informacijsko ozadje tako, da bi bil sposoben razumevati določeno specifiko industrijske problematike.
Glavna ovira za odkrivanje ključnih problemov domače industrije je se vedno avtocenzura, strah pred akcijo in seveda pomanjkanje določene civilnosti in demokratičnosti v gospodarski organizaciji. Ta demokratičnost se seveda ne nanaša na karkoli, temveč naj bi bila predmet strokovnega in kvalificiranega posluha vodilnih, ki poudarjajo svojo odgovornost. Dokler industrijskih okolij ne bomo civilizirali, tj. odprli za strokovno, organizacijsko, poslovno in predvsem kadrovsko kritiko in za civilne (korektne) odnose med delavci, vse dotlej bo vtecena nesposobnost gospodarila, rovarila in ovirala bistveno učinkovitost (kvaliteto in kvantiteto) proizvodnje.
V Delu, 18. avgusta 1988, se je B. Gruban neutemeljeno spotaknil ob brezbarvno in enostransko razpravo na drugem FI, ki ga je malomarno preimenoval v Forum informaticus. Mislim, da skrajno posplošujoči prispevki v Delu, ki povzdigujejo samozadostnost talentov brez upoštevanja možnosti za npr. industrijsko realizacijo talenta, uveljavljajo predvsem nedoločno namero pisca, ki ni navedel konkretnih ovir in blokad, ki se postavljajo na pot mladim in starim talentom kjerkoli in ne samo na področju Ril. "Znanje za razvoj" v Delu bi na področju Ril velikokrat zasluzilo kar naslov "Neznanje za razvoj".
A. P. Zeleznikar
□OPRTO PISMO DIALOGA?
NAPOVED ODPRTEGA STROKOVNEGA
Z dokajšnjim razburjenjem in nekaj prikrite klanske pripadnosti je bilo v različnih krogih, ki se ukvarjajo s informatiko in računalništvom, možno spremljati reakcije na Odprto pismo uredniškemu odboru časopisa In-formatica. A priori sera mnenja, da so odprta pisma dobronamerna in zato dobrodošla, zato tudi to potezo šestih bralcev jemljem kot željo in prispevek k spemembi obstoječega stanja na bolje. Verjetno se vsi zavedamo, da je tudi na področju računalništva in informatike obstoječe stanje v družbi nezadovoljivo. Utesnjenost, ki zaradi te nezadovolj ivosti nastaja, udari na piano v najrazličnejših oblikah. Poglejmo, kaj so glavne pomanjkljivosti, ki jih na področju informatike in računalništva opažam in nato bomo poskušali ugotoviti povezavo in vlogo časopisa Informatica v tem stanju.
ZA INFORMATIKO SE SE NISMO ODLOČILI
KAJ NAJ ČASOPIS INFORMATICA OMOBOCA
Čeprav vsak dan govorimo o informacijski družbi, (ponekod Je o post i n-f ormac i j ski druibi
-	econimy o-f mind), pa trdim, da se naèa druiba nikoli ni zares odločila za in-formatiko in računalništvo. Stanje v Sloveniji pa kaie:
-	Družbene dejavnosti nimajo sektorjev, ki bi se profesionalno ukvarjali z informatiko:
♦	ni Republiškega komiteja za informatiko (imamo pa 10 drugih republiških komitejev)
♦	Gospodarska zbornica Slovenije nima Splošnega združenja za informatiko in računal ništvo (ima na primer združenje
■ lesarstva,	usnjarsko predelovalne
industrije, zavarovalstva itd, skupaj 21 združenj).
-	Na področju visokega šolstva ,nimamo Fakultete za informatiko in računalništvo.
-	Res je da imamo v uporabi več tisoč računalnikov po delovnih organizacijah, toda ti računalniki niso nikakor orodje informatike. Večinoma niso povezani	v integriran informacijski sistem delovne organizacije, ampak so namenjeni raznim skoraj izoliranim finančnim in računovodski m aplikacijam. Delovne organizacije so,večinoma še daleč od množične uporabe računalnikov v planiranju, spremljanju in obračunu proizvodnje - to pa lahko prinaša kvalitetna sredstva in kvaliteto dela.
-	Računalniška industrija je slabo izkoriščena in nepovezana, za kar je krivih veliko vzrokov. Ponavadi velike računalniške hiše nikoli niso videle svoje vloge v pisanju kvalitetne aplikativne programske opreme in izgradnji integriranih informacijskih sistemov, ampak raje v preprodajanju materialne opreme in zastopniški dejavnosti. Slaba volja, ki jo stresajo na nekaj uspešnih privatnikov ali malih firm, je znak nezrelosti in nesposobnosti velikih organizirati prodorne kadre v skupne projekte.
- V dobi informatike.
ko je znanje tržna
vrednota, nimamo Instituta za računalništvo (imamo 32 institutov med drugim za informatiko, za mlekarstvo, druž i ne , .itd.).
infor'matiko in inštitutov ali bi omed i c i nsko za načrtovanje
-	Vrhunski strokovnjaki s področja informatike in , računal ni štva so., največkrat ■ tujek v industrijskem okolju. Ambiciozni posa'mezni ki , ki v takem okolju še najdejo voljo do usposabljanja, ' velikokrat	opravljajo podiplomski študij na lastne stroške ali pa celo ilegalno - brez vednosti delovne organi zaci je, saj računajo, da bi jim magisterij ali doktorat na delovnem mestu utegnil celo škoditi, saj so večkrat priča negativni selekciji.
-	Raziskovalni projekti (Evropski, mednarodni) gredo Večinoma mimo-nas in opažamo tehnološki zaostanek, ki se iz leta v leto veča.
-	Industrijsko povezovanje podjetij pri razvoju, proizvodnji in trženju na področju računalniške industrije gre mimo nas, kar nam otežkoča vstop na svetovni trg in nas pušča zadaj v organizaciji in izvedbi masovne proizvodnje računalniške opreme in kvalitetne programske opreme.
V tej precej nezavidljivi situaciji obstaja torej časopis Informatica, ki naj poveže in informira strokovnjake, ki se ukvarjajo z računalništvom in informatiko pri nas. Kaj naj bi bile naloge tega časopisa vemo, strokovnjaki pa so večinoma nezadovoljni s stanjem informatike, čeprav moramo poudariti, da nekateri posamezniki, posebej raz i skoval ci, dosegajo lepe uspehe.
Casopis.naj omogoča, izvaja ali vzpodbuja:
-	PUBLICIRANJE STROKOVNIH ČLANKOV: Vemo, da strokovni svet komunicira pretežno v angleškem jeziku in ker naj bi bili prispevki mednarodno usmerjeni, naj bodo napisani v angleškem jeziku. Hkrati naj bi bilo publiciranje v časopisu Informatica priprava za obilico mladih kadrov, da se izvežbajo za pisanje mednarodno kvalitetnih prispevkov (zato tudi v angleškem jez i ku) .
-	INFORMIRANJE: Številne raziskovalne in razvojne aktivnosti na področju informatike in računal ni štva naj se skupno objavljajo in s tem se omogoči pretok informacij v našem prostoru.
-	PROPASIRANJE: V tako zaostajajoči družbi kot je naša, ■ kjer se tehnološki razkorak -med razvitimi in nami tako veča, mislim, da mora časopis Informatica tudi izvesti	vlogo propagandne narave, kjer pokaže posamezne pozitivne akcije. Mislim, da je to vlogo časopisa Informatica dobro razumel glavni urednik časopisa.
-	KRITIKO: Časopis Informatica se je verjetno premalo ukvarjal s kritiko. Mislim, da lahko k izboljšanju številnih problemov pripomorejo prav posamezniki., ki se profesionalno ukvarjajo z računalništvom.
-	POVEZOVANJE: - Časopis Informatica naj bi vzpodbujal,povezovanje različnih strok, ki se navezujejo . na informatiko ali lahko nanjo vplivajo. Tu mislim na. področje biologije (genetski inženiring, neuralne mreže), medicine, sociologije, pa tudi filozofije, saj so •nekatera vprašanja informatike, posebej paralelnosti procesov in umetne inteligence zelo zanimiva tudi s tega stališča. Na univerzah v tujini (posebej Stanford) so zelo zanimivi' mešani i nterdiscipli narni timi, ki zajemajo informatike, matematike, filozofe, biologe, sociologe in druge ter se ukvarjajo z načelnimi' problemi na primer o r epr ez entac i j i znanja' v možganih in računal niku, modelih sklepanja in drugo..
PREDLOGI ZA DODATNE AKTIVNOSTI
Predlagam, da naj časopis Informati.ca prevzame tudi številne dodatne aktivnosti, . ki lahko oživijo i.n izboljšajo stanje informatike pri nas. Tako naj prevzame:
-	IZBOR ■ NAJBOLJŠEGA RAZISKOVALCA iz področja računal ništva , in informatike pri nas (metodologija izbora in kriteriji presegajo cilje tega članka). Mogoče bi; ob dobro znanih kriterijih bilo umesno sestaviti rang lestvico vseh raziskovalcev.
-	LESTVICO	10	NAJBOLJŠIH	STUDENTOV računal ništva in informatike (to je koristno zaradi pravilnega usmerjanja mladih kadrov, saj vemo, da mnogi potonejo v ubi jajočem okolju z negativno selekcijo).
-	LESTVICO 5 FINANCND NAJUSPEŠNEJŠIH DELOVNIH ORSANIZACIJ IN PRIVATNIKOV s področja računalništva (podatke je verjetno mogoče dobiti preko davčne uprave, SDK ali statistike). Kriterij naj bi bil na primer ostanek dohodka na zaposlenega.
-	RAZPIS NABRADNIH RAZISKOVALNIH NALOG ZA STUDENTE - teme naj bodo iz področja računalništva in i n-f ormati ke, zasnovane zelo na Široko, tako, da jih lahko izvajajo Studenti Fakultete za elektrotehniko. Studenti FNT (matematiki, -fiziki,...), bilogi, medicinci, -f i 1 oz o-f i , . . .
-	Objavljanje nekaterih zanimivih in uspeSnih raziskovalnih, izobraževalnih ter proizvodnih programov iz tujine. V teh informacijah naj bi bile vključena tudi poročila, kaj so nekatere družbe (driave, velike mul tinacionalke) naredile na organi zaci jskem, pravnem, -finančnem in družbenem področju, da bi pospeSile razvoj in-formatike pri njih.
STROKOVNOST
Verjetna bi druStvo in časopis In-formatica pridobil na strokovnem ugledu, če bi poleg dosedanjih	aktivnosti	organiziral več
strokovnih predavanj in zopet gostil kakSen mednarodni simpozij. Mogoče bi bilo zanimivo tudi založništvo strokovne literature.
vejico. Nekaj stavkov je stilno popolnoma nedopustno napisanih. Tak primer je stavek: Zakaj In-formatica objavlja prispevke, ki se prikazujejo kot znanost, pa to niso? Ne glede na to, da je to tudi pesniSki način izražanja, kamor sodijo retorična vpraSanja, je vsakomur takoj jasno, da se prispevki ne morejo prikazovati, ker gre v takem primeru za raeta-f izi ko. Prispevki so v primeru objave v In-formatici članki, ki so resda mogoče napisani na magnetnem mediju in v končni fazi natiskani na papir in se prav nič ne prikazujejo. Avtor bi lahko stavek napisal takole: Zakaj so v Informatici objavljeni lažno znanstveni prispevki? Se v tem primeru je taka izjava nekompetentna, ker avtorji očividno nimajo razčiščenih pojmov o tem, kaj je znanost in kaj je lažna znanost, medtem ko celo sami priznavajo, da se na znanstveno fantastiko ne spoznajo.
Najprimernejši sklep tega odgovora na odprto pismo bi bilo znano apelovsko reklo:"Le čevlje sodi naj kopitar!". Ob Šestih avtorjih pisma pa sem prepričan, da je imel vsak od njih priložnost, da bi pismo slovnično in smiselno popravil, zato bo tako sestavljeno pismo lahko upravičeno povzročilo bumerangovske odgovore.
Mag. Rihard Piskar, dipl. ing.
Korak naprej je tudi akcija, naj časopis dobi ISSN identifikacijo.
Strokovni ugled bi si časopis pridobil tudi z vzpodbujanjem, objavljanjem ali celo izdelavo neodvisnih strokovnih ekspertiz, ki bi zajemale od predloga za celovit nacionalni program informatike do ocene posameznih programskih paketov.
Jasno pa je, da z vlogo, ki jo za časopis Informatica vidim in je po mojem mnenju edino možna v naSem okolju, postaja časopis Informatica res informativen in ne časopis vrhunske svetovne kvalitete. Vsakemu bralcu je jasno, da bo svetovno kvalitetne članke Sel iskat drugam (če mu bo to dostopno). TakSno je stanje danes.	Saša Prešern
A JOINED DISCOURSE OF SCIENCE FICTION IN COMPUTER SCIENCE (I)
Anton P. Zeleznikar Iskra Delta, Ljubljana
That everyone can learn to read will ruin in the long run not only writing, but thinking too.
Friedrich Nietzsche [-3] 67
0. Introduction
ODGOVOR NA ODPRTO PISMO
Vsak izobraženec ima določen stil izražanja do katerega ima neizpodbitno pravico, dokler je v skladu z jezikovnimi normami. Ko pa njegov ustvarjalni zagon in svojskost izražanja nista več v mejah jezikovnega standarda, se bralec lahko vpraSa, če je tudi vsebina avtorjevega . prispevka sploh vredna logične razčleni tve.
2e nagovor so avtorji odprtega pisma končali z dvopičjem,' čeprav je pravilno ločila klicaj. Verjetno so avtorji dobili idejo za uporabo dvopičja v angleSkem jeziku, pa Se tam se praviloma konča z vejico ali piko, glede na to ali gre za neformalni ali formalni nagovor. Iz sl.ovenskega pravopisa poznamo stvarna lastna imena, v katera sodi tudi "časopis Informatica" in ta se ne sklanjajo, tako da je v rodi Iniku pravilno "časopisa Inf or/nati ca " in ne "časopisa Informati ce". Lahko se seveda uporablja samo lastno ime Informatica Informatice.	Izgleda, kot
uporabljajo figure pesniškega mnogorečje "zaslediti videti" sodi v to figuro, pa še tu je pravilno, da se besedi ločita z
tudi z rodi Ini kom da avtorji jezika, kajti
Some of them would want to have the science rented to themselves and albeit they are only mediocre individuals, they behave as a scienti fic elite; for the sake of appearance they are very active, however, in the essence, they produce merely sterile reports to get or to preservi^ their position.*
Vinko Kambič [-2] 9
In the open letter fO] to the Journal Informatica, signed by elite Slovene computer-scientists, one of the main reproaches was that some papers published in Informatica belong to science fiction, misrepresented science, and phi losophy. . Yes, also to philosophy! The last should be one of the most disturbing facts, which are not acceptable by the scientifically
* Nekateri hoćejo imeti znanost v zakupu in čeravno so mnogokrat samo povprečneži, se tudi vedejo kot znanstvena elita; na videz so zelo aktivni, a v bistvu piSejo le sterilna poročila za dosego ali ohranitev položaja.
(and probably technologically) oriented spirits of foreign scientific journals in the domain of computer science and, of course, to the Scientific Spirit of the open letter signatories. So, let us take an occasional look at some non-scientific quotations, philosophical statement, a, and ghosts of fiction, taken from eminent computer science journals and books.
For the sake of clarity, we shall put quotations from various sources (e.g. eminent professional journals) being recognized as cata-strophic seient. ific information (this maybe understood in the sense of the Greek 'kata-strofe' with the meaning 'overthrowing' or 'tricking down') between the particular symbol of catastrophe (^(number) and the symbol of the end of quotation J [<reference number)]. Beside this, quotations will be printed in Italic. In some cases we shal 1 give short comments on the contents of quotations. Quotations .shal] not be Listed in any systematic order.
1. In the Realm of Artificial
The computer scientist knows, of course. But he certainly does not know exactly what he knows.
Paraphrasing Teilhard de Chardin
In general, AI (Artificial Intelligence) is a discipline, for which it cannot be said that it belongs to a traditionally solid scientific orientation. In many cases, AI has to do with beliefs, hopes, promises, predictions, goals, programs, methods, and misunderstandings which could be ingenuously classified as non-scientific, fictitious, unbelievable, etc. .To step a bit further into pure scientific belief, investigations of this kind should be forbidden, because science has to follow the clearly understood, methodologically legal, and abstractly beaten tracks of research. These tracks of beliefs could probably belong only to the realms of legal mathematics, pure logic, algorithmic computer science, hard-mathematically formalized and semaritically reduced cognitive and psychological models, etc. Nowadays, this way of non-thinking and exactly e.\-erc i s ing disciplinarity can certainly be scientifically productive, but commonsensically and technologically does not remain on the way of development of modern research and understanding of the living and artificial.
Ijl Almost everyone has an opinion about artificial intelligence (AI). Either it's the wave of the future, or it's Just, a public relations canard, It's a new programming technology or it's a new mindset - a way of looking at problems that no other discipline has. Or, perhaps the name is an oxymoron and the whole idea of AI is absurd.	«J [1]
Within the domain of AI, we should like to have commonsense reasoning of a living being in a scientific and formalized form to enable an adequate modeling by computational reasoning. However, it is not clear yet how the first kind
of reasoning could be projected (modeled) adequately to the second kind of reasoning.
<}2 .'We would perhaps be better off using the term "computational reasoning " to represent what we do. Equipped with insights from computer scienoe, we are attempting to look at protil ems of reasoning and rationality not addressed previously by disciplines other than psychology and philosophy.	>1* [ 1J
AI certainly belongs to disciplines which are not only purely scientific, as are for example piiilosophy, psychoJogy, cognitive science, neurophysiology, and information processing in living organisms. Wishes and beliefs governing AI can be comprehended merely as goals and speculations of possible research which have not at all entered into the strict scientific, traditionally rationalistic realm yet.
In AT we use the computational metaphor to help us find answers to questions such as:
-	How can an intelligent system anticipate the future and prepare for it, without wasting effort on irrelevant or extremely improbable events?
-	How can we make systems that function with only an impoverished, partial knowledge base in the way that, humans are- able to do?
-	How can we communicate successfully with computers in (natural) languages fraught' with ambiguity and vagueness?
-	How can multiple entities plan and cooperate to achieve goals that no individual could ever hope to?
IVhile we can look to human behavior for insight into these issues, our main goals are to understand the phenomena of intelligent activity and find ways to exploit that understandi ng in artifacts.	>I< [1]
The questions of tiie last "cata-strophe" are philosophical for science cannot answer satisfactorily yet. They are maybe on the way of scientific research, but for a pure scientist they may represent an inadmissible approach of exactly structured scientific methodologies. At this point, we can understand how science proceeds on its way of scientific research through philosophy, which opens new horizons and illuminates the way of scientific progressing.
l^i We are far from being able to answer these questions, or even to integrate what little we know about intelligent behavior.	5 [1]
2. Knowledge Representation and Reasoning
What is knowledge? A traditional answer is that knowledge is a form of Justified belief. . . . Beliefs can be false, and the truth may not be believed.
Fred I. Dretske [-1] 85
/j5 A widely recognized goal of artificial intelligence (AI) is the creation of artifacts that can emulate humans in their ability to reason symbolically, as exemplified in typical AI domains such as planning, natural language understanding, di agnosi s, and tutoring.
Currently most of this work is predicated on a belief that intelligent systems can he constructed from explicit, declarative knowledge bases, which in turn are operated on by general, formal reasoning mechanisms. This fundamental hypothesis of AI means that knowledge representation and reasoning - the study of formal ways of extracting information from symbolically represented knowledge - is of central importance to the field.	>I< [2]
To emulate humans in their ability to reason symbolically embraces the entire realm of human metaphysics as information, including the domains such as planning, natural language understanding, diagnosis, tutoring, etc. Saying that most of these activities are based on beliefs accentuates only the metaphysical nature of problems in question. A realistic, good, prospective, and valid science cannot be founded on metaphysical, i.e. individual or personal phenomenology, but has to be constituted by culturally accumulated and disciplined information, which guarantees the existence of a realistic and valid science through selectively extracted information.
While the claim that intel ligent behavior can arise out of the computational manipulati on of propositional symbalie structures is certainly open to argument, AT has not yet developed a serious alternative to this working hypo thesis. At the moment, this stance is responsible for the vast majority of work in the field ...	5 [2]
How can intelligence arise out of computational manipulation, if this manipulation does not concern the realm (and nature) of information out of which intelligence can arise as a symbolic structure of information? However, the hypothesis of /j6 gives some orientation of a possible research when the computational manipulation could be replaced by the informational one. In this case intelligence would arise out of informational manipulation which is understood as Informing of information [10] .
<j7 It should be pointed out that the view of possibility of mechanized intelligence is not new. It seems to have originated with Leibniz' (1646-1716) dream of a calculus of ideas, wherein truths could be determined by manipulating an "alphabet of thought" (characteristica universalis) in some combinatory way, much akin to the way numerical expressions are manipulated in Newton's calculus. The crucial thing about such manipulation is that it would be required to preserve truth, similar to the way that numerical manipulation preserves value. >I< [ 2 ]
An alphabet of thought (characteristica universalis), as imagined by Leibniz, can be comprehended in the modern epoch as a kind of logic, say, informational logic TU, 12]. But within such a logic the preservation of truth could only be one of the commonsense aspects, which characteristically belongs to the doctrine of traditional logic. As pointed out in ijll (Hintikka), besides the true state of affairs, in the realm of informational, there are a number of other possible states of affairs which could be preserved. The most
general principle of preserving could be the principle of information (or principle of cybernetics), irrespective of the informational nat.uj-e (truth, possibility, temporalness, etc.).
ij8 , Our primary need is for a knowledge-based system to be able to realize all of the things that the sentences in its KB [knowledge base] determi ne to be true about the world. In other words, when asked a question, a, then the system needs to determi ne whether or not KB f= or ("h" means "entails" ) . ... Rather, what is taken into account is only the form of the two premises, along with a rule of inference that states that, if it is given V.v p(x) ^ q(x) ("for any x, if x has property p, then it has property q"), and it is given that p(a) for some particular a ("a has property p"), then it is acceptable to conclude that q(a) holds. We can say that conclusion is derivable from the given premises, and write
V.v p{x) ^ q(x), pia) I- q{a)
(the symbol "f-" means "derives") .
5 [2]
The question is how the sentences in a KB determine the truth about the world. And why only the truth of the world could be relevant? Why not put the question of how information concerning belief, possibility, awareness, necessity, reasoning, understanding or any other informational form or informational process could be preserved? Can really everything be reduced to the problem of truth? Why only truth may sound scientific? Etc.
/j9 Since proofs are Just a matter of form, and not content, it would seem that they are Just the kind of symbol manipulation that computers are good at. Tf we could show this to certain, and could tie the notion of derivation directly to the notion of entailment, then we would have a way for a computer to mechanically produce facts implicit in a KB. Indeed, there are two crucial relationships between entailments and derivations that close the loop for us : soundness and completeness theorems for predicate logic tell us that for a set of sentences, KB,
KB a if and only if KB h a
ÌS [2]
Deductive proofs are just a matter of form, and not content, and computers may be good at such a symbol manipulation.
<jlO ... the hope was that applications such as weather forecasting, pattern recognition, electronic fault diagnosis and disease identification could all be handled by single software systems.
This research was a failure : researchers set themselves too ambitious targets.	5S [4]
The coinment to the last "cata-strophe" would be that scientific research can be a failure. So, failures occur within scientific research, irrespective of the strictness of scientific disciplinarity. Certainly, ■ this happens more often in sciences whose principles of doctrinairism are still being developed, i.e. in younger and so-called "life" sciences.
Jjll The Japanese seizure of expert systems as a med i urn for their expansion in information technology resulted in a massive explosi on in funding. However, in Japan, interest seems to be fa Iteri ng. There are n number of i nd i rat i ons of this.
First, good staff are moving from the nationally funded loot projects back to their companies. Second, few Japanese researchers have published any worthwhile research results: major Journals on ai and software engineering are conspicuous by the total absence of any Japanese reports.	•!< [4]
It seems that throvigVi time i n s t i tu t i o.n a 1 funding of research can fail not only in Japan, but also here and now. Have our institutionally organized and funded signatories of the open letter [0] ever asked themselves which way they are moving scientifically, if they have not published any worthwhile research results yet?
<jl2 To build knowledge bases with thousands of rules is a major undertaking, and the resultant expert systems tend to be unreliable and very inefficient. Unfortunately, there has been little funding in this country for establishing the theoretical foundations which might help developers overcome this problem.	>1« [4]
Little funding could mean little theoretical foundations for knowledge bases with thousands of rules. Beside this, the present day expert systems are evident products of the so-called artifact builders who do not follow the principles of how the human cognitive mind expertizes problems.
/jl4 There has long been interest in both philosophy and AS in finding natural semantics for logics of knowledge and belief. The standard approach has been the so-called possible-worlds mndei. The intuitive idea, which goes back to Hintikka ... , is that besides the true state of affairs, there are a number of other possible states of affairs, or possible worlds. Some of these possible worlds may be indistinguishable from the true world to an agent. An agent is then said to know or believe fact cp if <p is true in all the worlds he thinks possible.	5 [5]
The last "cata-strophe" stresses the problem of uncertainty, where in the domain of possible worlds not only the true state of affairs is imaginable. Here the philosophy and AI are searching together, trying to find possible semantics for logics concerning different informational phenomena as are knowledge, belief, awareness, reasoning, etc. A strict separation between philosophy as thinking and AI as science is not anymore possible, at least not in the domain of the research of mind which is becoming as philosophical as scientific. In this connection, the injuriousness of an interdisciplinary or common research cannot be at all defended commonsensical1y.
1^15 Unfortunately, in real life people are certainly not omniscient. Indeed, possible-worlds advocates have always stressed that this style of semantics assumes an "ideal" rational reasoner, with infinite computational power. But for many appiica ti ons, one would like a logic that provides a more real i s t i c repi-esentati on of human reasoning.	5 [5]
3. Belief, Awareness, and Reasoning
Belief, awareness, and reasoning, as rational notions, may belong only to non-scientific or to weak scientific categories. For each of these it is necessary to develop a particular logic, or only one universal logic, which could be adequately particularized for specific cases. The author believes that informational logic [11, 12] can be particularized or universalized to any needed or application-adequate extent.
^jlS Several new logics for belief and knowledge are introduced and studied, all of which have the property that agents are not logically omniscient. Tn particular,. in these logics, the set of beliefs of an agent does not necessarily contain all valid formulas. Thus, these logics are more suitable than traditional logics for modeling beliefs of humans (or machines) with limited reasoning capabili ti es.	J [5]
To be not logically omniscient means to be logically specialized or particularized. Thus, it is possible to construct separate logics for describing beliefs, awarenesses, and reasonings. Each of th(?se logics is characterized by special logical operators, which fit particular notions or semantics. So, we are. on the way to search for universal logic, which in particular cases could be induced to the needed particularity.
It should be said that the protagonists of prepositional logic have the intention to make this logic omniscient in the sense of mastering wi. th its reduced means any possible phenomenology. On the other hand, in many applications one would like to have a more adequate or realistic logic, closer to the human way of reasoning.
ijlB Another interesting direction to take is that of considering quantified versions of these logics. Here some very interesting technical and philosophical questions arise. For example, since we would like to be able to capture sentences such as "He is aware of something that I am not aware of," we seem to be forced into allowing states with different domains, and dealing with all the technical complications that arise there. There is still much work to be done in finding an intuitively motivated logic powerful enough to describe such situations.	ij [5]
Article [51 begins with the quotation of Teilhard de Chardin: "The animal knows, of course. But it certainly does not know that it knows." This quotation is similar to that "He is aware of something that I am not aware of", from	In [6], the author described de
Chardin's statement by the formula a . In
this formula a is the the so-called animal's
metaphysics (an animal's total information), f^é
marks "certainly does not know" and	denotes
K
"knows". In general, the meaning of this formula would be that information a is not
informed that it informs. In the second case we can put the following equivalences:
Y	('he' = 'his metaphysics')
"Df
'someth i ng'
a	('T' = 'my own metaphysios'
f=a =Df	aware',
=
Df
'is not aware'
M	1=	a ;
t=	M	a :
¥		a :
t=	a	
	a	h:
	a	bs:
fei	a	
a	N	)=:
a		N:
<x		
Now the formula which describes the second case is, for example,
Already in [7] several cases of this kind of statements are formally described:
1= 1= a: a is informed that it is informed;
a is not informed that it is informed; a is informed that it is not informed; a is not. informed that it is not i n formed ;
a is informed that it informs; a is not informed that it informs; a is informed that it does not inform; a is not informed that it does not inform;
a informs that it informs; a does not inform that it informs; a informs that it does not inform; <x	a does not inform that it does not
inform.
In these formulae, the particular!zation (substitution) of the operators )= and can be non-uniform. For the first formula, for instance, |=.j, t=g « has the meaning, a is
informed true that a is believed, etc.
<jl7 Two remarkable trends in automation can be seen. One is " i ntel 1 ectual i za t i on " and the other is "systcmatization ". The former aims to improve the performance of machines to the level of a human operator.	>5 [8]
To improve the performance of a today's machine to the level of a human operator may sound unrealistic or at least non-scientific. However, the intention of scientific research has to remain in the domain of unknown, otherwise it cannot follow the aim of any research: to reveal the unrevealed.
1)18 The computer is superior to the human brain in memory, information retrieval and logical reasoning, but inferior in intuition, macroscopic judgment, commonsense knowledge penetration, creation, adaptation, flexible thinking and so on. In a word, the computer can perform the part of the left hemisphere of brain but not the right. Therefore, we should reconsider the role of the computer as a part of the man-machine system..	>5 [8]
That computer can perform the part of the left hemisphere of brain but not the right may sound as a psychological paraphrase but not as a legal scientific statement. Besides, this statement is very ambitious even if it is said that computer can perform only a part (wh.at functions of the part) of the left hemisphere
activities. Yes, this could be a computer from a science fiction story. But, it is certainly also true too that ma ny performances of computers are comparable to the functions of human mind.
<jl9. . .. special commonsense knowledge is a metarule which covers the deficit, harmonizes the contradiction, and increases the adaptability and flexibili ty of the special knowledge. 5 [8]
Some readers would agree that the last "catastrophe" is philosophical and can hardly be brought into a legal scientific context.
4. Philosophy Concerning Computer Science
'j20 Epistemologi cal issues have been a concern of iifestern philosophy at least as far back as the Sophists ... . Cognitive science, on the other hand, is still in its childhood as an interdisciplinary pursuit which attempts to integrate the resources of artificial intelligence, cognitive psychology, philosoohy, and perhaps other areas. ... it is now time for epistemology to begin to reap the benefits of cognitive science.	>X< [9]
Cognitive science, as described in the last "cata-strophe", as a field of interdisciplinary research essentially concerns AI and through it the entire domain of the computer science. Psychology and philosophy influence the research in AI as well as in computer science quite substantially. Due to specific requirements of the AI field, especially particularized philosophies evolve, which may speed up the development of a new scientific and technological understanding.
/321 Further, Stefik and Bobrow belong to a predomi nant tradition within AI that sees little value in foundational (called "philosophical" ) questions. Shortcomings of current techniques and systems are seen as a transient failure to cover the right details and to invent the right clever mechanisms, rather than as anything more basic. Instead of radical challenge to foundations, they would rather see arguments over the technical details ... They imply that if we focus on the details of the technology, somehow it will all work out. They indulge in sci enee-fict i on speculations, such as "... never computer systems ... might acquire background that ue humans bring to bear to understands the world classes of complexity for computational beings that relate in natural ways to Piaget's stages novels of Isaac Asimov. This kind of innocent optimism about technological achievement has been common among AI researchers since the beginning of the discipline.	5 [13]
Obviously, the signatories of the open letter [0] also belong to a predominant and additionally subcultura] tradition within the domestic computer science, that finds no value in philosophical qijestions by which foundations could be widened. They cannot see anything more basic, but they believe, that the present day shortcomings in computer science are a transient failure, which will vanish through time when traditional means will be developed
in those details which now imply failures. And right this kind of concluding let them speculate in the domain of science fiction by the same sentences as quoted in the last "catastrophe" .
intelligent mechanistic items. But how can a hard, scientifically declared discipline be other than mythica], if one of the most relevant goals is the preservation of unreal beliefs and of solid professional organization (organicismi?
5. Of AI Scientists
Computer scientists are becoming more and more AI scientists. They are the creators and explorers of that what we may call computer science and artificial intelligence. Thus, they have their own way of understanding what the science in which they create may illuminate, enlighten, or bring to the surface of their own consciousness. But, there is a substantial difference in thinking between computer scientists and AT scientists: the first are more like hard-workers, dwelling in a computer-specific hard and soft doctrines; they walk through a computer land like somnambulists; the second are like discoverers, breeding soft and disciplinary open doctrines. But, there are also .subs t. an t i a] differences in the i nf ormat i onfi Ì env i ronnn^n t.s of t.he first and the second ones.
ij22 There are two types of AI scientists ; call them P and not-P. Here is a simple test to tell which type you are. You are type not-P if you immediate!y close a book containing sentences like this: "We are led, accordi ng to Descartes, to an agent that circumvents the hegemony of the mechanical-detei-ministic nexus by virtue of its immateriality and freedom ..." If you are type not-P, skip this book; you wiJi not miss any important new clues about writing intelligent programs. Type P readers, however, will find it a useful and at times illuminating discussion of mayor concepts in the philosophica 1 and psychological analysis of mind.	.J [3]
The signatories of the open letter [0] belong obviously to the not-P scientists. The advantage of P scientists is that they can read and find useful to get insight into the philosophical and psychological analysis of mind. If nothing else, the not-P scientists remain narrowed in their "fach" doctrines, and utilitarian in respect to any philosophical enlightenment which might come from new ideas, concepts, methodologies, etc. The not-P scientists are the real guardians of their scientific disciplines, keeping the piirity and unchangeabi1 ity of their disciplines under the severe control. T t is evident that in scientific research the not-P scientist will in principle not support a new, innovative research achievement.
liZ3 As it is, it is still one of the best of the recent attempts to find, explain or construct the appropriate philosophical and psychological context for AI, /4s such it is one of the few hooks available on the mythical subject of cognitive science.	5(3]
IjZA The rise of neural computing can be seen as part or the i nterplay between two schools of thought in ai: the cognitists and the artifact buiIders.
The artifact builders are interested in building software systems which exhibit intelligent behaviour and do not concern themselves with whether the underlying mechanisms of the software resemble the way humans operate or not. The cognitists are interested in using the computer as a laboratory for exploring notions about the way human think.	>X< [4]
In some way, the not-P and P scientists can be compared to artifact builders and cognitists, respectively. There exists some congeniality between a not-P scientist and an artifact builder. The first one builds his science i~r(!S])oc t i ve of n relevant philosophy and the second one constructs his artifacts as useful tools or even as plausible toys. On the other hand, a P scientist and a cognitist is very sensible to philosophy (e.g. Descartes, Pascal, Leibniz, Hilbert, Rüssel, Wittgenstein, etc.), so he can innovate the existing science by bringing a new semantics into the scientific context.
6. Some Comments to the First Part
The intention of listed quotations and comments to them in this part of the discourse was to show how some listed quotations can be understood and misunderstood as science, science fiction, or misinterpreted science. This is the way of information development in the field we call scientific research. In the area of research, beliefs, awarenesses, and reasonings follow wishes, hopes, and cares for the future, survival, but also for individual scientific progress, individual curiosity, and certainly the will for power. Scientific world is in no way free from personal aspirations, narrow interests, and ideologies of knowledge. In the history of science this kind of dilemma appears as a regular one. "Every triumphant theory passes through three stages: first it is dismissed as untrue ..." (embryologist Von Bear). In this context, the signatories of the open letter [0] may appear as the guardians of what has been said at the beginning of Introduction: some of them would want to have the science rented to themselves and ...
References
No human information is completely free of mythical beliefs. Not only the cognitive science, but also AI and computer science include myths in believ ing that intelligent objects can be constructed through non-
[-3] F. Nietzsche: Thus Spoke Zarathustra. Penguin Books (Reprinted 1985).
[-2] V. Kambie: in Knowledge for Development
(in Slovene). Delo, Aug 11, 1988, Ljubljana.
[-1] F.I. Dretske: Knowledge and the Flaw of Information. Basil Blackwell Publisher, Oxford (1981 ) .
[0]	T. Pisanski & All: An Open Letter to the Editorial Board of the Journal Informatica (in Slovene). Informatica 12 (1988), No. 3, 78-79.
[1]	R.J. Brachman, F.H. Henig: The Emergence of Artificial Intelligence Technology. AT&T Technical Journal 67 (1988), Issue 1, 3-6.
[2]	R.J. Brachman: The Basics of Knowledge Representation and Reasoning. AT&T Technical. Journal 67 (1988), Tssuf; 1, 7-21.
[3] O.J. Revi ews Lindsay). 385.
Flanagan: The Science of Mind. Book and Response (Reviewed by R.K. Artificial Intelligence 34 (1988),
[4]	D. Ince: .4 New Dawn Breaks in Japan as Research Sinks in the West. Computing (March 17, 1988) 18-19.
[5]	R. Fagin, J.Y. Halpern: Belief, Awareness, and Limited Reasoning. Artificial Intelligence 34 (1988) 39-76.
[6]	A.P. Zeleznikar: Information and Information System (in Slovene). Proceedings of the Conference on Information Systems, 7-16, Technical Faculty, Maribor, June 22-24, 1988.
[7]	A.P. Zeleznikar: Information qua Information. Proceedings of MIPRO '88, New Generation Computers, 4/15-30.
[8]	T. Terano: Fuzzy Logic in Man-Machine Systems. New Generation Computing 5 (1987) 131132.
|9| S.W. Sraoliur: A Review of A.I. Goldman, Rpistemology and Cognition. Artificial Intelligence 34 (1988) 251-267.
[10] A.P. Zeleznikar: Information. Cybernetica 31
Pri nei pi es 1988) 99-122.
of
[11]	A.P. Zeleznikar: Informational Logic I. Informatica 12 (1988), No. 3, 26-38.
[12]	A.P. Zeleznikar: Informational Logic II. Informatica 12 (1988), No. 4,3-20.
[13]	T. Winograd, F. Flores: On Understanding Computers and Cognition : A New Foundation for Design. A response to the reviews. Artificial Intelligence 31 (1987) 250-261.
O KECKNZIJAH
Razumljivo je, da so recenzije o istem delu lahko nekoliko različne, pač odvisno od recenzentovega natančnega vpogleda v področje, ki ga recenzira, od njegove osebnosti, pa tudi od njegovega trenutnega razpoloženja. Težje pa razumemo bistveno različnost recenzij istega
dela. Kot primer bi nanizal recenzijo prispevka, ki je bil poslan na jugoslovanski mednarodni simpozij (priloga 1) in odgovor urednika ameriške revije (priloga 2), kamor je bil poslan isti prispevek v prevedeni obliki. Sodbo o enakosti vsebin recenzije in odgovora si ob branju ustvarite sami,
(I.R.)
(priloga 1)
D. ftko se rad ne prlln/aća, niollmo obrazIoSenJe za autora: Uvod, zaključek in priložene tri slike dajejo vtis nekakšne potegavščine, katere smisel pri najboljši volji nisem mogel dojeti. Mislim da bi vodstvo aimpozbija moralo odločno protestirati pri avtorjevi delovni organizaciji, oziroma opozoriti to organizacijo, da ji njeni člani rušijo strokovni ugled.
(priloga 2)
I bel i eve that you have an in ter est ing idea -for an article here, but 1 would like to see more details of the language and more description D+ what it is used for (is it -for Computei—ftided Design, Computei—Aided Manufacturing, or what?) There are questions about some oi th^j^hings contained in it that might be cleared up by a more ex tensive expos i t ion. For example, you state "s ing le ca 11 rule: every subroutine can be called once only." I-f that were strictly true, it would be more economical to code the subroutines in-line; what I assume you mean is that subroutines are not re-entrant — there can be at most one invocation of the subroutine at any given time.
The artwork is interesting, and I would be happy to publish it together with an expanded article, IF you could bear in mind that the standard format on this side of the ocean is 8.5" x 11" <28 cm x 21.5 cm). Usual ly the foreign submissions that I rece i ve have marg ins that can be-irimmed; that is not the case on the drawings.
If you care to resubmit the article, or any others, I will be happy to consider them. I don ' t know whether you are a member of SIGSMALL/PC or not; ifnot, and you don ' t get the newsletter, I will say that I have rewritten your previous contributions for the sake of grammar and spelling; I hope that this causes you no distress.
it
H. K. Hodge Editor, SIGSMALL/PC Notes
A SHORT COMMENT TO ENGLISH READERS
The upper Enclosure 1 (priloga 1) is a Slovene referee's report to the program committee of a Yugoslav computer conference. This Enclosure has to be compared to Enclosure 2 (priloga 2) which is the editor's of SIGMALLS/PC Notes opinion on the same paper. The English translation of Enclosure 1 is the following:
D. If the paper is not accepted we would like to have an explanation with arguments for the author :
The introduction, conclusion, and the three drawings enclosed give the impression of a kind of fraud, which to my best will, I could not comprehend. I think that the committee of the symposium must resolutely express a protest at the author's working organization or let this organization know that its employees destroy its professional reputation.
(The translator regrets the inconvenient diction' of the translation of the Slovene original into English which is typical for the herostratic style of the referee's opinion. By this diction the spirits of the referee's opinion remains true.)	APZ
SE O RECENZIJAH
A. P. Zeleznikar, Iskra Delta
Ker gre v prispevku "O recenzijah", v katerem je priloženo tudi dokazno gradivo, za značilno domačo afero, naj mi bo dovoljeno, da dodam svoje skromno mnenje kar v domačem jeziku. Takoj na začetku ponavljam vprašanja o histeričnem pamfletiranju, rotenju, zaklinjanju, herostratstvu oziroma o motivativnih konfliktih starejše generacije, ki sem jih načel v [1], Kako se obraniti teh strastnih in čustvenih (in mestoma že kar sovražnih) izlivov in njihove škodljivosti pri prodiranju novih akterjev oziroma njihovih izvirnih dosežkov na domačih prizoriščih v okviru izrazite znanstvene subkulturnosti? Ali bomo morali začeti resneje in konkretneje postavljati vprašanja, kdo so ti naši, nam neznani avtoritativni usmerjevalci v znanosti, ki povzročajo zaradi svoje preživele strokovnosti blokade tam, kjer jih domala nihče več ne pričakuje. Kot da smo priča nekemu razvojnemu rasizmu, ki nam omejuje in predpisuje spontan in s svetom usklajen razvoj domače znanstvene misli. Seveda nikakor ne trdim, da so pojavi recenzentskega rasizma prisotni samo na področju računalništva in informatike, zasledimo jih lahko tudi pri fizikih, zlasti v segmentih intenzivnega in filozofsko pogojenega znanstvenega razvoja.
Upam seveda, da gre v primeru navedene recenzije res za osamljen pojav, katerega nosilec in izvajalec ima v širšem jugoslovanskem okolju le obrobni (slovensko
značilni , su bk u 1 1. u r-n i ) \'f)! i v. 'l'f> pa seveda odvezuje odgovornosti organizatorjev strokovnih srećanj, za katere se takärie recenzije pišejo. Ker, kakšni pa bodo post.a li doina<''i simpo/. iji in posvetovanja, če bo njihova strokovna vsebina odvisna od ljudi, ki nastopajo kot čuvarji balkanskega kndoksn? Njim bi vpljni preprosti nasvet, da v svoj« roiMiri/.en tsk<! tjarnit.uri; ne vključujejo kogarkoli, zlasti pa ne pri vr^.itnrev herostratstvn, ki bi bili sposobni zadušiti redke pojave domače znanstvene in tehnoloSke ustvarjalnosti.
Seveda bi bila tudi moja reakcija na omenjeno recenzijo zgolj čustvena, če ne bi znal argumentirati in po svoji vesti interpretirati vsebine žaljive recenzije. Organizator posvetovanja pričakuje pod D korektno strokovno obrazložitev zavrnitve, ki bo sporočena avtorju. Pričujoča zavrnitev pa ni le napad na integriteto avtorja, temveč je še poziv na sankcijo proti ugledu njegove delovne organizacije. To pa je hkrati argument, ki ga lahko izkoristi nekdo drug, recenzentskemu herostratu podoben herostrat, npr. pri visokošolski oziroma strokovni. reelekciji avtorja. 'Strokovne utemeljitve zavrnitve prispevka v recenziji seveda ni, je pa motivativni konflikt, ki se izraža v izjavi, da je prispevek informacijsko tako oddaljen od informacijskega ozadja recenzenta, da ga ta ne more več dojemati pri vsej svoji volji za razumevanje prispevka. Prav tu pa nastane vprašanje strokovne usposobljenosti recenzenta in njegove normalnosti: kakšno je njegovo informacijsko strokovno ozadje, da prispevka ne more več dojemati in ga prav zaradi te svoje nesposobnosti imenuje potegavščina. Očitno tak recenzent nima več kaj iskati v strokovnem informacijskem prostoru, v katerem deluje ne le kot recenzent temveč kot akter stroke, raziskovalec, znanstvenik, pedagog, inženir in naposled tudi kot človek.
k (Ml k r<i t M i. doli pi'ispevka, ki so utemeljeno nesprejemljivi, napačni rezultati, nerazumljiv jezik, konkietna poprava in dodelava in kar je Se podobnih trdnih kriterijev. Veliko tega pa je pokazanega prav v prilogi 2, kjer si je avtorju popolnoma tuj urednik vzel prizadevno toliko časa, da mu svetuje in pomaga prek oblikovnih in vsebinskih težav. V tem se razpoznava konstruktivni mentorski odnos in odgovornost recenzenta do avtorja. Seveda se recenzent Lahko postavi tudi na stališče, da ocenjuje posplošujoče takole: znanstvena fantastika, lažiznanost, čudne izjave, s stališča recenzenta nerazumljivo, sramotno, skrb zbujajoče, strokovno rušilno, neznanstveno itd. Toda prav to se od strokovnega recenzenta ne pričakuje. Pomislite, kako bi lahko razsojalo sodišče, ki bi uporabljalo posplošujoče kriterije na nedoločen in volutari s t ičen način. V tem primeru stroka ne bi bila kaj več od političnega voluntarizma.
Nazadnje bi seveda lahko postavil še vprašanje prizivnosti avtorja in njegove delovne organizacije. Avtor je del svojega priziva prav gol.ovo izrazil z objavo dokumentov. Vendar je tu zasluga za njegovo pozitivno obrambo zgolj njegova in se lahko zaradi njegove lastne iznajdljivosti dobro konča tudi zanj. Mislim pa, da bi morala avtorjeva delovna organizacija izraziti tehten pomislek jugoslovanskim organizatorjem posvetovanja, ki so spregledali nestrokovnost in (nezavedno) zlonamernost recenzenta, predvsem pa nedopustno mešanje strokovnih in zasebnih (konfliktnih) podmen. Ali to pomeni, da balkanski strokovni simpoziji blokirajo informacijo prav tam, kjer je ta mednarodno relevantna? Ce recenzenti ne morejo več pri svoji najboljši volji razumevati razvojnih tokov "svoje" stroke po svetu, potem pač učinkovito Sirijo svojo strokovno nesposobnost povsem nočeš-hočeš na perspektivne domače avtorje.
Prepričan sem, da je gornja osvetlitev vsebine recenzije zadostna; sedaj pa bi bilo mogoče še povedati, kaj bi v negativni recenziji pričakoval kot pojasnilo za avtorja. Argument za zavrnitev bi lahko bil; neoriginalnost, premajhna izvirnost, neupoštevanje standardov stroke.
Slovstvo
[1] A.P. Zeleznikar: Se o soncih v praznem prostoru. Informatica 12 (1988), št. 3, 81-83.