18 KINESIOLOGIASLOVENICA3 (1997)1: 18-23 
Frane Erčuli * 
EVALU4llON OF THE EXPERT SYSTEM »KISS 
1.2 KOSZ1 EKSPERT« ON PLAYERS OF THE 
SLOVENIAN CADET NATIONAL FEMALE 
BASKETBAll TEAM 
PREVERJANJE VELJAVtJOSTI EKSPERTNEGA 
SISTEMA »KISS 1.2 KOSZ1 EKSPERT« NA 
VZORCU ŽENSKE KADETSKE KOŠARKARSKE 
REPREZENTANCE SLOVENIJE 
ABSTRACT 
The model of the expert system of potential perfor-
mance was evaluated ona sample of twelve female 
basketba ll players of the Slovene National Cadet 
Team. A battery of twentyfive morphological and 
fourteen motor tests was carried out immediately 
before the players' departure far the European 
Championships w hich enabled us to establish the ef-
fect of the present moment sta te of preparedness on 
the playing performance and efficiency of the play-
ers. 
In order to establ ish the relationship between man-
ifest (predictive) variables and the criterion (playing 
performancev~nd efficiency), the expert system 
»KISS 1.2 KOSZ1 EKSPERT«was used . 
The estimate (assessment) of playing performance is 
in statistically sign ificant correlation with the expert 
evaluation of potential performance on the highest 
(ultimate) level of performance tree (O.S 7), and with 
expert evaluation of potential performance on the 
level of morphological dimensions (0.59). The cor-
relation between playing efficiency and expert esti-
mate of potential performance is just under the lim-
it of statistical significance. 
With guards and forwards the amount of correlation 
between different criteria of success and expert eval-
uation of potential performance does not differ sig-
nificantly. The values range from 0.43 to 0.69, but 
due to the very small number of subjects, are not sta-
tistically significant. 
Key words: basketba/1, women, cadets, motorics, 
morphology, expert system, validation 
IZVLEČEK 
Na vzorcu dvanajstih košarkaric slovenske kadetske 
reprezentance, smo ovrednotili model ekspertnega 
sistema potencia lne uspešnosti. Igralke smo, 
neposredno pred odhodom na evropsko prvenstvo, 
izmerili z baterijo petindvajsetih morfoloških in štiri-
najstih motoričnih testov. Na ta način smo ugotavljali 
kako trenutna pripravljenost igralk vpliva na nj ihovo 
igralno uspešnost in učinkovitost. 
Za ugotavljanje povezanosti sistema manifestnih 
(pred iktorskih) spremenljivk s kriterijem (igralno 
uspešnostjo in učinkoviS<2stjo) smo uporabili ekspert-
ni sistem »KISS 1.2 KOSZ1 EKSPERT«. 
Ocena uspešnosti igranja je v statisti čno značiln i 
povezavi z ekspertno oceno potencialne uspešnosti 
na najvišjem (končnem) nivoju drevesa uspešnosti 
(0,57) in z ekspertno oceno potencialne uspešnosti 
na nivoju morfoloških razsežnosti (0,59). Povezanost 
med igralno učinkovitostjo in ekspertno oceno po-
tencialne uspešnosti je malo pod mejo statistične 
· značiln osti. 
Pri branilkah in krilnih igralkah so rezultati po-
vezanosti med različnim i kriteriji uspešnosti in 
ekspertno oceno potencialne uspešnosti dokaj 
izenačeni. Vrednosti se gibljejo med 0,43 in 0,69, 
vendar zaradi zelo majhnega števi la merjenk niso 
statistično značilne. Povezanost spremenljivk mor-
fološkega in motoričnega prostora s kriteriji upešnos-
ti je pri branilkah dokaj uravnotežena, pri krilnih 
igralkah pa so morfološke razsežnosti v večji 
povezanosti s kriteriji igralne uspešnosti in učinkovi­
tosti. 
Ključne besede: košarka, kadetinje, reprezentanca, 
morfologija, motorika, ekspertni sistem, veljavnost 
Frane Erčulj 
EVALUATION OF THE EXPERT SYSTEM »KISS 1 .2 KOŠŽ1 EKSPERT« ON PLA YERS OF THE SLOVENIAN CADET .. . 19 
INTRODUCTION 
Basketball is a polystructural complex sport (13). It is 
a rather complicated and complex team bali game 
of which cyclic and acyclic movingstructures as well 
as contrary i nterests of the opponent teams are char-
acteristic. 
The performance of a basketball team or individual 
male or female players depends ona great number 
of factors having different effects on playing perfor-
mance. Most of the factors that have effect on the 
achievement i. e. performance in play (game) are 
more or less related to each other, they can comple-
ment, compensate orexclude each other. lndividual 
contribution is not directly measurable, the result 
(achievement) does not depend solely on the per-
formance of individual players but also on the co-
operation of the players which can be direct or indi-
rect (playingtactics in offense and in defense) as well 
as on the play of opponent players. Players create 
and adapt their movements with regard to the play-
ing situation, thus also intuition plays an important 
role. 
Owing to unpredictability, great possibility of com-
pensation and multidimensionality, research work 
in basketball is very complicated and also less reli-
able. 
Because of the above, investigating the theory of per-
formance in basketball demands a systematic ap-
proach. The system of p laying performance of a 
player (basketball team) can only function if it com-
prises al l the essential elements of the system and 
their relationships (15). 
Theoretical knowledge of the factors on which play-
ing performance is dependent is one of the basic 
conditions necessary for a proper and successfu l 
training process in basketbal l. In this way only a 
coach-kinesiologist is able to conduct and control 
training process and to achieve rationality so indis-
pensable for top performance in modem sport. 
Playing performance isto a high degree affected by 
motor and morphological dimensions (2, 3, 9) there-
fore in this study major emphasis will be la id upon 
the investigation of this segment of the psychoso-
matic status. 
The use of artificial intelligence in research has in-
creased recently. Also kinesiological science has en-
riched its methodology by usi ng the methods, tech-
n irn 1P~ ;:mrl tnnk nf r1rtifir ir1 I intPlliPPnrP (17) 
Expert systems also belong to computer programs of 
artificia l intelligence. Expert systems behave similar-
ly to an expert in a particular field, their basic char-
acteristic being the ability to explain the solution. A 
computer helps to collect and arrange a segment of 
our knowledge and to employ this knowledge in a 
consequent, systematic and transparent way (16). 
Expert systems are being more and more and with 
increasing success used in the investigation of the 
theory of sport performa nce. The method applied 
to research complex systems combines computer 
capacities with the experts' knowledge in the field 
of sport. A basketball player or a basketball team can 
also be defined asa (very) complex system (14). 
Expert systems asa rule consist of three modu les (11 , 
16, 17): 
1.Knowledge base 
It is based on the sports science findings on the 
preparation of athletes. In it the knowledge consists 
of generally approved facts and heuristics (expert 
rules of inference and decision making). The perfor-
mance model must be designed so that it can give 
answers to the fol lowing questions: 
- on which factors, abilities and characteristics the 
result (achievement) depends 
- how significant each of the factors is and what its 
relative share in the result is 
- whatthe relationships between single factors are. 
The quality of the expert system isto the highest de-
gree dependent on the amount and quality of the 
knowledge base. 
2. Mechan isms of lnference 
It enables active application of the knowledge from 
the knowledge base for problem solving. It gives 
answers to the user's questions and explains them. It 
determines the sequence of steps for the problem 
sol uti on. 
3. User's (communication) lnterface 
It takes ca re of comfortable commun ication be-
tween the user and the system and allons the user 
the insight into the process of problem solving di-
rected by the mechanism of inference. 
The first to begin the investigation of potential and 
competitive (playing) performance in basketbal l (by 
means of different models of expert sytems), was 
Dežman (4, 5, 6, 7). He also composed a simplified 
and generalized model of the tree of potential per-
fnrm;,nr,=, n f;, h;,d,,:,th;, II nl::1\/Pr 17\ 
20 Frane Erčulj EVALUATION OF THE EXPERT SYSTEM »KISS 1.2 KOŠŽ1 EKSPERT« ON PLAYERS OFTHE SLOVENIAN CADET .. 
lndividual factors (d imensions) have different effects 
on the playing performance, therefore their contri-
bution i. e. weights are determined by decision mak-
ing rules. Decision making rules are of qualitative 
character and are determined by an expert. They il-
lustrate the relationship between the elements of the 
performance model. The experts expresses the 
equation of specification according to his own con-
ception and arbitrary judgement (17). 
By determin ing normalizers the level of single di-
mensions is evaluated on the tree of performance 
with values 1 to 5. Normal izers are of quantitative 
character and are used to set limits to the resu lts in 
single variables. Decision making rules and normal-
izers can be determined for ali types of basketball 
players. 
On the basis of information obtained from literature 
and so far existing studies in th is field (4, 5, 6, 10, 11, 
17) and of expert knowledge, decisional making 
rules (weight) were determined in motor and mor-
phological space of the psychosomatic status of the 
young female basketball players (9). Decision mak-
ing rules were made for each playing posit ion. The 
reduced model ofthe expertsystem of potE;_n_tial per-
formance was named »KISS 1.2 KOSZMOT1 
EKSPERT«. 
Potential performance in the motor abi lity space was 
defined by 14 tests, by means of which condit ional 
abilities, technical knowledge and coordination abil-
ities were measured. Amongconditional abilities are 
included power, velocity and endurance, technical 
knowledge and coordination abili ties are deter-
mined with movements without and w ith a bali. 
Morphological space was defined by tests measur-
ing external geometrical dimensions (longitudinal, 
transversal and voluminosity dimensions) and inter-
na! geometrical d imensions (fat tissue). 
The purpose of the evaluation of the model of ex-
pert system described isto obtain as objective infor-
mation as possible on the state of preparedness of 
the young female basketball players and thus as ob-
jective as possible prognostic assessment of their 
competitive performance. This is also one of basic 
issues in the training process in basketball and in 
sport in general. 
METHODS 
The sample was comprised of 12 players of the 
Slovene National Cadet Female Basketball Team 
preparing for the participation at the European 
Championship, 1995. In further analysis 11 players 
w ith sufficient playing t irne during the competition 
were included if they met the follow ing criteria: 
- that they were bom in t he years 1978, 1979 or 
1980 
- that they were measured with all tests 
- thatthey were in good health and free of injures of 
locomotion apparatus during measurement and 
competition. 
With regard to thei r playing positions and roles in 
play, the selected sample was d ivided into two sub-
samples: outside p layers or guards (n=5) were sep-
arated from the players whose p laying position is 
nearer to the basket i. e. inside players or forwards 
(n = 6). Th is division was done by the coaches of the 
team. 
The sample of (independent) variables of potential 
performance comprised the tests applied in the 
model of tree of expert system of prognostic perfor-
mance, i. e. the tests measuring those motor and 
morphological dimensions with the highest degree 
of impact on playing performance of young basket-
ball players (see Tables 2 and 3). Very good measur-
ing characteristics of t hese tests have al lready been 
proven severa! t imes (1, 2, 9). Detailed description of 
the tests was given by Erčulj (9). 
The sample of dependant variables was comprised 
of variables estimating playing (competitive) perfor-
mance and efficiency: 
- Playing performance of the young players was as-
sessed subjectively: the coach and his assistant 
evaluated the player's performance w ith the val-
ues from 1 to 5. Globa! playing performance, i. e. 
total performance of the player in offense and de-
fense was assessed accord ing to the cri teria by 
Erčulj (9). 
- The efficiency of the individual player in the game 
was stated objectively by the calculation of effi-
ciency indexes (5). The parameters necessary for 
the calculation of indexes (conclud ing offensive 
and defensive actions) were obtained from offi-
cial game statistics. 
The measurements were carried outat the premises 
of the Faculty of Sport immed iately before the de-
parture to the European Cadet Championship which 
ena bled us to establ ish the effect of the state of pre-
paredness on the players' competitive performance. 
Competitive performance was assessed at the 
European Cadet Championship held in August 1995 
at Cetniewo (Polanci). 
The data was processed at the Institute of 
Kinesiology at the Faculty of Sport in Ljubljana, on 
Frane Erčulj 
EVALUATI0N 0 F THE EXPERT SYSTEM »KISS 1.2 K0ŠŽ1 EKSPERT« ON PLAYERS 0FTHE SL0VENIAN CADET .. 21 
the VAX/VMS computer of the University Com-
puting Center in Ljubljana. Statistical program pack-
age SPSS was used. The expertsystem was made by 
the help of computer program KISS (version 1 .2 E) 
which was made for the needs of Basketball 
lnformation System by Leskošek B. (12). 
The model of the expert system (KISS 1.2 KOŠŽ1 
EKSPERT) and descriptive statistics procedures were 
used for data processing. The correlation between 
final estimates of potential and competitive model 
of performance were obta ined by means of Pearson 
correlation coefficients. 
RESULTS 
The procedures described were primarily used to 
measure playing performance and efficiency for the 
players with a sufficient amount of playing t irne at 
the championship. Table 1 shows the results of play-
ing performance and efficiency for above men-
t ioned players. 
Table 1 : Playing performance and efficiency of the 
players at the European Cadet Cham-
pionship 1995. 
Pt pp CEPP AIPE RIPEN GAMfS ,PTIME 
1 G 3 7.75 0.88 4 12.9 
2 F 2 3.00 0 .69 3 4.0 
3 G 2 9.86 0.94 7 26.9 
4 F 3 5.60 0.98 s 13.8 
s F 4 16.71 1.1 3 7 38.1 
6 F 2 3.86 0.73 7 13.1 
7 F 3 3.20 0.81 5 9.8 
8 G 3 18.86 1 .18 7 34.6 
9 G 2 14.71 1.05 7 26.1 
10 F 1 8.1 4 0.92 7 16.8 
11 G 1 o o 2 3.4 
Legend: 
PL - player 
PP - playing position 
CEPP - coach's evaluation of playing performance 
A IPE - absolute index of playingefficiency at the competition 
RI PE - re lative i ndex of playing efficiency at the competition 
GAM ES - number of played games at the tournament 
PTIME - average p laying tirne on the game (minutes) 
With the use of the expert system »KISS 1.2 KOŠŽ1 
EKSPERT« and by means of a computer program me 
(KISS 1.2) the potential performance was computed 
for each of the players. Decision making rules and 
normalizers were used to evaluate the level of indi-
v idual d imensions of potential performance. Table 
2 shows expert evaluations of potential performance 
on the three highest levels of the performance deci-
sion tree. 
Table 2: Expertevaluations of potential performance 
of the cadet national team players on the 
three highest levels of the performance tree. 
J;, Pl,ifuqi; 
1 
2 
3 
4 
s 
6 
7 
8 
9 
10 
11 
Legend: 
FINAL 
MORPH 
MOTOR 
ChtiiL. PP, FINAL MORPH MOTOR 
G 2.7 1.8 3.3 
F 2.7 2.6 2.8 
G 3.8 3.9 3.7 
F 3.3 2.5 4.1 
F 3.1 4.0 2.2 
F 3.0 2.2 3.8 
F 2.7 3.2 2.2 
G 3.3 3.6 3.1 
G 2.8 2.8 2.8 
F 3.2 3.3 3 .1 
G 2.6 2.6 2.6 
expert evaluatio ns of potential performance on 
the h ighest level (final leve! o f the performance 
tree) 
expert evaluations of potential performance on 
the highest level of morphological d imensions 
expert evaluations of potential performance on 
the h ighest level of motor dimensions 
The validity of the results of the used expert system 
was established with the amount of correlation be-
tween criteria variables (variables of playing perfor-
mance and effi ciency- CEPP, AIPE, RIPE) and the 
variables of potential performance on the three high-
est levels of the performance tree (FINAL, MORPH, 
MOTOR). 
Tables 4 show the results of val idity of the used ex-
pert system also for the sub-samples of guards and 
forwards. 
DISCUSSION 
lf the sample of players is regarded asa whole, the 
correlation between potential estimate of perfor-
mance on the highest level (FINAL) and the playing 
performance (CEPP) is statistically significa nt. The 
criteria of objectively established playing efficiency 
(A IPE, RI PE) are in relatively high correlation with the 
final estimate of potential performance (FINAL) but 
the val ues of correlation coefficients are not statisti-
22 Frane Erčulj EVALUATION OF THE EXPERT SYSTEM uKISS 1.2 KOŠŽ1 EKSPERT« ON PLAYERS OF THE SLOVENIAN CADET ... 
Table 3: Correlation between the criteria variables and the variables of potential performance on the three 
highest levels of the performance tree. 
0.81 1.00 
0.83 0.79 1.,00 
0.67 0. 70 0.77 1.00 
0.87 0.95 0.76 0 .78 1.00 
0.57 0.43 0.52 0 .64 0.58 1.00 
0.59 0.58 0.41 0.54 0.70 0.60 1.00 
0.04 -0.11 0.1 5 0.20 -0.07 0.55 -0.33 1.00 
*statisticaly significant (P<0.05) are values, higher than O.S 7 (marked with emphasised print) 
Table 4: Correlation between the criteria variables and the variables of potential performance on the three 
highest levels of the performance tree for guards (G) and forwards (F) separately. 
0.89(G) 1.00 
0.86(F) 
0.93(G) 0.92(G) 1.00 
0.88(F) 0.88(F) 
0.99(G) 0.89(G) 0.89(G) 1.00 
0.29(F) 0 .56(F) O.SO(F) 
0.96(G) 0.95(G) 0.89(G) 0.96(G) 1.00 
0.86(F) 0.97(F) 0.87(F) 0.67(F) 
068(G) 0.43(G) 0.SO(G) 0.69(G) 0.69(G) 1.00 
0.47(F) 0.47(F) 0.67(F) 0.58(F) 0.48(F) 
0.6l(G) 0.46(G) 0.34(G) 0.66(G) 0.70(G) 0.88(G) 1.00 
0.58(F) 0.81 (F) 0.72(F) 0.36(F) 0.76(F) O 08(F) 
0.52(G) 0.25(G) 0.53(G) 0.48(G) 0.43(G) 0.79(G) 0.39(Gl 1.00 
-0.18(F) -0.38(F) -0.17(F) 0.07(F) -0.32(F) 0.57(F) -0. 77(F) 
• statisticaly significant (P<0.05): guards - 0.81, forwards - O. 75. 
cally significant. Rather interesting is the fact that the 
correlation between mean playingtime (PTIME) and 
expert estimate of potential performance is signifi-
cant (0.64). It can thus be concluded thatthe players 
with higher estimates of potential performance have 
a higher amount of playing tirne compared w ith 
those with lower estimates of potential performance. 
Playingtime is thus a kind of criterion of playing per-
formance since better players asa rule have a high-
er amount of playing tirne than less successfu l play-
ers, and vice versa. 
Among dimensions on the highest level only the cor-
relation between playing performance and mor-
phological characteristics is statistically significant, 
the correlation between motor dimensions and play-
ing performance of the National Team players is, due 
to extremely low values of correlation coefficients, 
practically negl igible. This probably does not mean 
t hat morphological dimensions could have greater 
influence on players' performance than motor di-
mensions. It can neither be concluded that the play-
ers' motor dimensions are less developed. It is pos-
sible thatthe players do not differ significantly in mo-
tor dimensions and that morphological dimensions 
are responsible for differentiation. These relation-
ships undoubtedly demand further investigation. 
These findings are on ly valid for the selected sample 
of players, as the sample, although of high quality, is 
too small and does not perm it generalization. 
 
Frane Erčulj 
EVALUATION OF THE EXPERT SYSTEM »KISS 1 .2 KOŠŽ1 EKSPERT« ON PLA YERS OF THE SLOVENIAN CADET .. 23 
Coefficients of correlation between d ifferent crite-
ria of performance and the fina l estimate of potential 
performance are rather the same with both sub-sam-
ples and range from 0.43 to 0.69. As the sample of 
five (5) and six (6) players is too small, they are not 
significant. 
The correlation between potential performance of 
guards and forwards and the amount of playing tirne 
is relatively high (0.69 and 0.48) and is a good indi-
cator of playing performance, i. e. of the rank of a 
player, but not statistical ly significant. 
lf the correlation between morphological and mo-
tor dimensions and playing performance is consid-
ered separately it is found to be rather similar w ith 
guards. With forwards the correlation between mor-
phological dimensions and performance criteria is 
much higher. 
1 n spite of th is a relatively h igh degree of correlation 
between the model of potential performance of the 
players and their playing performance and efficien-
cy can be established. With regard to the fact that in 
th is investigation a reduced model of expert system 
was applied and only a part of psychosomatic status 
of the players was involved, a relatively high level of 
predictive value of expertsystem was established on 
the sample of the National Female Cadet Basketball 
Team. 
REFERENCES 
1. Dežman B. Zanesljivost in faktorska veljavnost testov os-
novne in košarkarske motorike (Reliability and factor validi-
ty of basic and basketball motor tests). Ljubljana: Fakulteta 
za telesno ku lturo, Inštitut za šport, 1983. 
2. Dežman B. Določanje homogenih skupin na osnovi neka-
terih antropometričnih in motoričnih razsežnosti pri mladih 
košarkarjih (Determining homogeneous groups of young 
basketbal I players on the basi s of some anthropometric and 
motor dimensions (doctoral dissertation)). Ljubljana: 
Fakulteta za telesno kulturo, 1988. 
3. Dežman B. Povezanost nekaterih bistvenih razsežnosti ra-
zličnih prostorov psihosomatičnega statusa mladih košarkar-
jev z uspešnostjo igranja (The correlation of some principal 
dimensions of various subspaces of the psychosomatic sta-
tus with playing success of young basketball players). 
Ljubljana: Fakulteta za telesno kulturo, Inštitut za kinezi-
ologijo, 1989. 
4. Dežman B., Jošt B. Spremljanje uspešnosti športnikov na os-
novi ekspertnega sistema (M onitoring athletes' performance 
on the basis of an expert system). Ljubljana: Športna zveza 
Slovenije, 1991. 
5. Dežman B. Ekspertensystem -Model zur Erfolgsprognose der 
Spieler im Basketball. Proceedings of the 6th ICHPER -
Europe congress: Praga, 1992: 111-117. 
6. Dežman B., Leskošek B. Ekspertni sistem ocenjevanja 
nadarjenosti otrok za igranje košarke (An Expert System for 
Assessment of Talent for Basketball in Children). 
Proceedings of the II. lnternationa l symposium »Sport of 
the young«: Ljubljana - Bled, 1993: 40-46. 
7. Dežman B., Erčulj F. Die Anwendbarkeit des 
Expertensystem-Modell zur Erfolgsprognose junger 
Basketballsp ieler. Proceedings of the lnternational 
Conference on »Science in sports team games«: Biala 
Podlaska, 1995: 187-194. 
8. Dežman B. An example of usefulness of expert modelling in 
sport. Proceedings of the international symposium »Sport 
motorics«: Praga, 1995. 
9. Erčulj F. Ovrednotenje modela ekspertnega sistema po-
tencialne in tekmovalne uspešnost i mladih košarkaric 
(Evaluation of the model of expert system of potential and 
competitive successfulness of young female basketball play-
ers). Ljubljana: Fakulteta za šport, 1996. 
1 O. Erčulj F., Dežman B. Unterschiedliche anthropometrische 
und motorische dimensionen bei 13- und 14- j(hrigen bas-
ketballspielerinnen, die auf verschiedenen spielpositionen 
spilen. Proceedings of the lnternational Conference on 
»Science in sports team games«: Biala Podlaska, 1995: 216 
- 223. 
11. Jošt B., Dežman B., Pustovrh J. Vrednotenje modela 
uspešnost i v posameznih športnih panogah na podlagi 
ekspertnega modeli ranja (Evaluation of a model of success 
in d ifferent sports on the basis of expert modelling). 
Ljubljana: Fakulteta za šport, Inštitut za kineziologijo, 1992. 
12. Leskošek B. KISS 1.2 E (Computer programme). Ljubljana: 
Fakulteta za šport, 1995. 
13. Matvejev L. P. Grundlagen des sportlichen Trainings. 
Berlin: Sportverlag, 1981. 
14. Mulej M. et al. Teorije sistemov (System theory). Maribor: 
Ekonomsko-poslovna fakulteta, 1992. 
15. Pavlovič M. Raziskovalni model v košarki (A research mod-
el in basketball). Trener- košarka, 1983; 19 (2): 13-17. 
16. Rajkovič V., Bohanec M., Šturm J., Leskošek B. An expert 
system for adcising children in choosing sports. 
Proceedings of the l. lnternational symposium »Sport of the 
young«: Ljubljana - Bled, 1990: 641 -646. 
1 7. Šturm J. et al. Izbor in usmerjanje otrok v športne panoge na 
podlagi ekspertnega modeliranja (lnitial Selection and 
Orientation of Ch ildren to Engage in Sports on the Basis of 
ExpertModelling). Ljubljana: Fakulteta za šport, Inštitut za 
šport, 1992.