https://doi.org/10.31449/inf.v46i1.3776 Informatica 46 (2022) 95–100 95 
Intelligent Course Recommendation Based on Neural Network for 
Innovation and Entrepreneurship Education of College Students 
Jinding Zou 
E-mail: 00194026@wzu.edu.cn 
School of Design and Art, Wenzhou University of Technology, Wenzhou, Zhejiang 325000, China 
Student paper 
Keywords: neural network, innovation and entrepreneurship education, recommendation technique, intelligence course, 
collaborative filtering 
Received: October 9, 2021 
This paper focuses on intelligence course recommendations for college students’ innovation and 
entrepreneurship education. Firstly, the traditional collaborative filtering algorithm was introduced, and 
then a new recommendation technique was designed based on an artificial neural network (ANN). The 
experimental data were collected through a crawler framework. The two methods were compared and 
analyzed. It was found that the training time of collaborative filtering and ANN was 16.78 s and 12.36 s, 
the testing time was 2.64 s and 2.12 s, the hit rate (HR) were 0.6078 and 0.6264, and the normalized 
discounted cumulative gain (NDCG) values were 0.2948 and 0.3356, respectively. The results reveal that 
ANN was more efficient in computation and better in recommendations. The results demonstrate the 
effectiveness of the ANN method for intelligent course recommendations. The method can be applied to 
the selection of innovation and entrepreneurship education courses for college students. 
Povzetek: Opisana je metoda nevronskih mrež za izbiro primernih študijskih predmetov. 
 
1 Introduction 
With the popularization of computers, the course selection 
system for college students has also changed. The original 
academic year system, in which all the courses were 
arranged by the school and completed according to the 
school’s course plan, has developed to a credit system, in 
which students choose courses independently based on 
minimum credits for graduation. With the development of 
network technology, the method of online course selection 
has been popularized, but the independent selection of 
courses often leads to unsatisfactory final grades and is not 
conducive to smooth learning because the selected courses 
are not strongly related to the courses they have already 
taken [1]. If the recommendation of courses can be 
achieved through intelligent methods [2] to help students 
choose suitable and interesting courses, it will be a good 
contribution to the education of college students. The 
recommendation technique refers to a method of targeted 
recommendation of information based on individual 
interests and behaviors, which has been widely used in e-
commerce [3], social entertainment [4], etc. Liu [5] 
designed a method based on an improved clustering 
algorithm to find users’ nearest neighbors before making 
recommendations by clustering users and items separately 
and found through theoretical analysis and experiments 
that the recommendation accuracy of the method has been 
significantly improved. He et al. [6] studied the tweet 
recommendation of SINA micro-blog, extracted user-
focused topics by k-cores analysis, used RS and linear 
regression to determine the parameters, and established a 
recommendation model based on semantic network. The 
experiment found that the method could make a timely and 
personalized recommendations for SINA micro-blog 
tweets. Tan et al. [7] designed a recommendation model 
based on probability matrix decomposition, which 
combined the social relationship and content information 
of items to improve the accuracy of recommendations. 
The experiment showed that the method was scalable and 
had a good recommendation effect. Due to the difference 
between course recommendation and commodity 
recommendation, recommendation techniques in course 
selection have been less studied. Zhang et al. [8] proposed 
an improved algorithm based on the Apriori algorithm, 
which preprocessed the data through Hadoop and then 
mined association rules. Through a series of experiments, 
they found that the method had good efficiency and was 
more applicable to the current Massive Open Online 
Courses (MOOC) platform. Based on neural networks, 
this paper analyzed the course recommendation method 
for college students’ innovation and entrepreneurship 
education and verified the effectiveness of the method 
through experimental analysis. This work makes some 
contributions to promote the intelligent development of 
courses and also helps to promote the further development 
of college education. 

96 Informatica 46 (2022) 95–100  J. Zou 
2 Neural network-based 
recommendation technique 
2.1 Innovation and entrepreneurship 
education for college students 
Innovation and entrepreneurship education is a new 
education mode that combines various educational 
concepts such as innovation, entrepreneurship and quality, 
and its main content is to strengthen the innovation and 
entrepreneurship consciousness of college students, shape 
the innovation and entrepreneurship spirit, improve the 
innovation and entrepreneurship ability, and cultivate the 
innovation and entrepreneurship morality, so that college 
students can achieve flexible employment and 
independent entrepreneurship. Its main purposes are as 
follows. 
(1) In line with the trend of the times: With the 
development of society, encouraging entrepreneurship, 
supporting entrepreneurship and establishing an 
innovative country have become key content. The 
employment pressure has become more and more 
prominent since the enrollment expansion of colleges. In 
order to ensure that college students can be successfully 
employed, it is necessary to carry out innovation and 
entrepreneurship education and create a better 
employment environment and more employment 
opportunities to improve the country’s innovation ability. 
(2) Cultivating innovative and entrepreneurial talents: 
With the progress of society, the competition among 
countries is also reflected in the competition of innovative 
and entrepreneurial talents. Innovation and 
entrepreneurship education is conducive to implementing 
the strategy of developing the country through science and 
education and promoting the progress of the country and 
society. 
(3) Promoting higher education reform: Conducting 
innovation and entrepreneurship education can bring new 
vitality and content to the education reform and is also a 
good way to improve the employability and psychological 
quality of college students, which can support the current 
higher education reform. 
(4) Promoting the overall development of college 
students: Innovation and entrepreneurship education can 
encourage students to actively innovate, help them better 
understand the laws of the market, seize the opportunity to 
develop from passive employment to active employment 
and from active employment to active entrepreneurship, 
and maximize the development of personal creativity. 
In the current curriculum of college students, different 
innovation and entrepreneurship education courses are 
arranged, and students can choose the corresponding 
courses to study according to their majors and needs. In 
order to better play the role of innovation and 
entrepreneurship education courses, this paper adopts a 
neural network-based method to recommend courses 
intelligently to students. 
2.2 Collaborative filtering technology 
The collaborative filtering technique is a relatively mature 
and widely used method in recommendation technology 
[9], which can recommend items to users according to 
their needs [10]. Its principle is as follows. For user A and 
the set of items that the user is interested in, the set of items 
is analyzed to find user B that is most similar to user A. 
Then, the items in the set of items of user B that have not 
been selected by user A are recommended to A. This 
method is applied to the intelligent course 
recommendation for college students’ innovation and 
entrepreneurship education. The process can be described 
as follows. 
(1) Establishing a matrix for student course evaluation: 
The matrix is established by students’ evaluation of a 
course. For student 1, his evaluation of course1 is C 11; then, 
student M’s evaluation of course N is C mn. The established 
matrix is shown in Table 1. 
 
 Course 
1 
Course 
2 
...... Course N 
Student 1 C 11 C 12 …… C 1n 
Student 2 C 21 C 22 …… C 2n 
...... …… ……  …… 
Student 
M 
C m1 C m2 …… C mn 
Table 1: Student course evaluation matrix. 
(2) Selecting students’ nearest neighbor: The 
similarity between students is calculated based on the 
above matrix. The corresponding formula is: 
Sim(u
a
,u
b
)=
∑ (C
a,i
−C
a
̅̅̅̅
)(C
b,i
−C
b
̅̅̅̅
)
i∈I
a.b
√∑ (C
a,i
−C
a
̅̅̅̅
)
2
i∈I
a.b
√∑ (C
b,i
−C
b
̅̅̅̅
)
2
i∈I
a.b
, 
where i refers to the courses evaluated by students a 
and b, C
a,i
  and C
b,i
  refer to the common evaluation of 
course i by students a and B, C
a
̅̅̅
  and C
b
̅̅̅
  refer to the 
average score of the evaluation. By calculating and 
ranking this similarity, k students with the most similarity 
are selected to establish students’ nearest neighbor, 𝑈 𝑠 . 
(3) Recommending courses intelligently: After 
getting the nearest neighbor 𝑈 𝑠 , the scores of the courses 
are predicted, and the formula is: 
𝑃 𝑢 ,𝑗 =𝐶 𝑢 ̅̅̅
+
∑ 𝑆𝑖𝑚 (𝑢 ,𝑢 𝑘 )(𝐶 𝑢 𝑘 ,𝑖 −𝐶 𝑢 𝑘 ̅̅̅̅̅̅
)
𝑢 𝑘 ∈𝑈 𝑠 ∑ (|𝑆𝑖𝑚 (𝑢 ,𝑢 𝑘 )|)
𝑢 𝑘 ∈𝑈 𝑠 , 
where  𝐶 𝑢 ̅̅̅
  refers to the average score of all courses 
given by student u, 𝐶 𝑢 𝑘 ,𝑖 refers to the score of course i, and 
𝐶 𝑢 𝑘 ̅̅̅̅̅
 refers to the average score of all courses. 
Through calculating and sorting 𝑃 𝑢 ,𝑗 , the first N 
courses are recommended to students as a set to complete 
the intelligence course recommendation for college 
students’ innovation and entrepreneurship education. 

Intelligent Course Recommendation Based on... Informatica 46 (2022) 95–100 97 
2.3 Neural network-based 
recommendation technique 
In this paper, an intelligent course recommendation 
technique for college students’ innovation and 
entrepreneurship education was designed based on 
artificial neural networks (ANN). ANN is a method that 
simulates how the human brain nerve processes 
information to solve problems [11]. ANN is characterized 
by a strong self-learning capability that enables the 
algorithm to achieve higher accuracy [12]. It has an 
extensive application in fields such as face recognition 
[13]. The algorithm first performs weighted summation on 
all the signals before processing the previous input, and 
the calculation formula is: 
𝑢 𝑖 =∑𝑤 𝑖𝑗
𝑥 𝑗 ̅ 𝜃 . 
When the input 𝑢 𝑖 of the neuron >0, it indicates that 
the neuron is in an excited state; otherwise, it is in an 
inhibited state. Then, the output of the neuron is obtained 
by performing an activation operation on the input, and the 
calculation formula is: 
𝑦 𝑖 =𝐾 (𝑢 𝑖 ), 
where K(∗) refers to the activation function. 
The idea of ANN is applied to intelligent course 
recommendations for innovation and entrepreneurship 
education. The weight of college students’ course is 
calculated according to the idea of ANN. Suppose that 
college students have n innovation and entrepreneurship 
education courses to choose from, and the weight of each 
course can be written as: 𝑊 𝑖 (𝑖 =1,2,⋯,𝑛 ), each course 
has m(m=0,1,⋯,m) leading courses, and the influence 
coefficient of every leading course is 𝑢 𝑖 =
1
𝑚 . ff the 
student has not browsed the course once, it makes the 
weight of the course +1. Each time a student views the 
course, the weight of the course is increased by 1. ff a 
student has taken t leading courses of some course, then 
the influence weight of t leading courses is added to the 
weight of the course is: 
W
I
=∑ u
j
w
pre
t
j−1
, 
where 𝑤 𝑝𝑟𝑒 refers to the weight of the leading course. 
After calculation, k courses that rank high in terms of 
weight are recommended to students as a set. 
3 Experimental analysis 
Since there is a lack of public datasets about course 
recommendations, this paper crawled the data needed for 
the experiment from the MOOC platform of Chinese 
universities through a crawler. In the MOOC, many 
colleges offer courses about innovation and 
entrepreneurship education, and some of them are shown 
in Table 2. 
It was seen from Table 2 that the innovation and 
entrepreneurship education courses offered on MOOC 
were different in terms of target audience and professional 
focus. Some courses were more inclined to cultivate 
innovation consciousness, while some were more inclined 
to improve entrepreneurial ability; therefore, intelligent 
course recommendation was needed to help college 
students better choose the appropriate course. 
The Uniform Resource Locator (URL) address of the 
detail page of every innovation and entrepreneurship 
education course was crawled through the Scrapy crawler 
framework. The crawled content included user identity, 
course scores, etc. The courses without scores and users 
who scored less than ten courses were excluded. The 
crawled data were cleaned. 70% of the data were randomly 
selected as the training set, and 30% were as the test set. 
The data sets are shown in Table 3. 
The experimental device was a MacBook Pro with 
Intel Core i7 2.2GHz processor, 16 G memory, 251 G hard 
disk, and Python 2.7 software environment. 70% of the 
obtained dataset was randomly selected as the training set, 
and the remaining 30% was used as the test set. 
The evaluation indicators of the algorithm are as 
follows. 
(1) Hit rate (HR): it was used to describe whether the 
recommendation results of the algorithm were in the top k 
recommendation lists, and its expression is: 
HitRatio@K=
Number of Hits@K
|GT|
×100%, 
where |𝐺𝑇 |  refers to the number of test sets and 
Number of Hits@K  is the sum of the number of the 
recommendation results that are consistent with the 
predicted results of the test set. 
(2) Normalized discounted cumulative gain (NDCG) 
[14]: it was used to describe the difference between the 
predicted and true results, and its expression is: 
{
 
 
 
 
 
 
𝐶𝐺
𝑘 =∑ 𝑟𝑒𝑙𝑖 𝑘 𝑖 =1
𝐷𝐶𝐺 𝑘 =∑
2
𝑟𝑒𝑙𝑖 −1
𝑙𝑜𝑔 2
(𝑖 +1)
𝑘 𝑖 =1
𝐼𝐷𝐶𝐺 𝑘 =∑
1
𝑙𝑜𝑔 2
1+𝑖 𝑘 𝑖 =1
𝑁𝐷𝐶𝐺 𝑘 =
𝐷𝐶𝐺 𝑘 𝐼𝐷𝐶𝐺 𝑘 , 
where 𝑟𝑒𝑙𝑖 refers to the relevance of the 
recommended results (if 𝑟𝑒𝑙𝑖 =1, then there was a hit; if 
𝑟𝑒𝑙𝑖 =0, then there was no hit), 𝐶𝐺
𝑘 is the cumulative 
gain, 𝐷𝐶𝐺 𝑘 is the discount cumulative gain, and 𝐼𝐷𝐶𝐺 𝑘 is 
the ideal discount cumulative gain. The larger the value of 
NDCG was, the better the recommendation was, i.e., the 
higher the overlap between the recommendation list given 
by the model and the real list. 
The running time of the two algorithms was first 
compared, and the results are shown in Figure 1. 
It was seen from Figure 1 that the training time of the 
ANN algorithm was 26.34% less than that of the 
collaborative filtering algorithm (16.78 s vs. 12.36 s) and 
the test time of the ANN algorithm was 19.7% less than 
that of the collaborative filtering algorithm (2.64 s vs. 2.12 
s). The results revealed that the ANN algorithm showed 
better computational efficiency in both training and test 
and had better availability in the recommendation of 
innovation and entrepreneurship education courses. 
 

98 Informatica 46 (2022) 95–100  J. Zou 
Course 
name 
Instituti-
ons 
Course introduction 
Chinese 
medicine 
innovation 
and 
entrepre-
neurship 
Chengdu 
University 
of 
Traditional 
Chinese 
Medicine 
Innovation and 
entrepreneurship are combined 
with traditional Chinese 
medicine health. The main 
content is how to grasp the 
good opportunity of 
entrepreneurship in developing 
the traditional Chinese 
medicine health industry. 
Innovation 
and 
entrepre-
neurship 
practice 
Shandong 
Transport 
Vocational 
College 
Based on the new trend of 
entrepreneurship and the 
characteristics of the college 
student group, the course 
focuses on improving the 
quality of college students’ 
entrepreneurship and solving 
common problems in different 
stages of the entrepreneurial 
process, which helps students 
to use innovative thinking to 
solve problems encountered in 
the process of future 
entrepreneurship and improve 
the comprehensive quality and 
ability to start and manage 
enterprises. 
Design 
thinking 
and 
innovative 
design 
Zhejiang 
University 
The research results in the field 
of innovative design in China 
are combined with cutting-edge 
achievements in design 
thinking in China and abroad. 
The course aims to effectively 
enhance product innovation and 
entrepreneurial practice. 
Innovation 
and 
entrepre-
neurship 
manage-
ment 
Nanjing 
University 
of Posts 
and 
Telecomm
unications 
The course helps students 
understand the rules and 
characteristics of 
entrepreneurship, recognize the 
problems they may encounter 
in the process of starting a 
business, draw lessons from 
them, and improve 
competitiveness. 
Innovation 
manage-
ment 
Zhejiang 
University 
The course helps students to 
develop their creative abilities 
and acquire relevant theoretical 
knowledge and practical skills. 
Foundati-
ons of 
college 
student 
entreprene
urship 
Wenzhou 
University 
The course helps students 
master the basic theory and 
process of entrepreneurship, 
establish a correct view of 
innovation and 
entrepreneurship, and guide 
them from theory to practice. 
Table 2: Examples of innovative entrepreneurship 
education courses. 
 Training set Test set 
Number of users 3650 1564 
Number of courses 1300 557 
Number of ratings 319398 136885 
Table 3Experimental data set. 
 
Figure 1: Running time comparison. 
Then, the HR and NDCG of the two algorithms were 
compared, and the results are shown in Figure 2. 
 
 
Figure 2: Performance comparison between algorithms. 
It was seen from Figure 2 that the HR of the ANN 
algorithm was 3.06% higher than that of the collaborative 
filtering algorithm (0.6078 vs. 0.6264) and the NCDG 
value of the ANN algorithm was 13.84% higher than that 
of the collaborative filtering algorithm (0.2948 vs. 0.3356). 
The results revealed that the ANN algorithm showed 
better performance in both HR and NDCG values; thus, it 
had better reliability in recommending innovation and 
entrepreneurship education courses for college students. 
4 Discussion 
The Internet has become more and more widely used in 
people’s lives. fn order to effectively obtain the required 
information from the increasing network information and 
resources, “information retrieval” and “information 
filtering” technologies emerged. The former refers to 
retrieve the required information with keywords and other 
methods [15], and the latter refers to filter the information 
on the network with users’ personal information to 
recommend information to users. For example, 

Intelligent Course Recommendation Based on... Informatica 46 (2022) 95–100 99 
recommendation systems [16] solves the problem of 
identifying relevant resources from a large number of 
available choices [17], and the collaborative filtering 
algorithm is the most commonly used one [18]; however, 
during courses selection, due to the small differences 
between users, i.e., students, in age and major and the fact 
that the number of courses is not as huge as the e-
commerce recommendation system, it is not applicable to 
intelligent course recommendation. 
Innovation is the inexhaustible power of development. 
Strengthening innovation and entrepreneurship education 
for college students has become a key content in the 
college curriculum [19], which is also a necessary way to 
cultivate innovative and entrepreneurial talents needed by 
society. With the development of network technology and 
the popularity of online learning, innovation and 
entrepreneurship education courses can also be learned 
online, which has a positive effect on improving the 
innovation consciousness and innovation ability of college 
students [20]. Therefore, this paper mainly analyzed the 
intelligent course recommendation for innovation and 
entrepreneurship education of college students. 
The comparison between the traditional collaborative 
filtering algorithm and the neural network-based 
recommendation technique suggested that the ANN 
algorithm had better performance in terms of both the 
running time and the recommendation performance. 
Figures 2 and 3 showed that the training time and test time 
of the ANN algorithm were 12.36 s and 2.12 s on the same 
data set, i.e., its computational efficiency was higher than 
that of the traditional collaborative filtering algorithm, and 
the HR and NDCG values of the ANN algorithm were 
0.6264 and 0.3356 respectively, i.e., the performance of 
the ANN algorithm was also superior to that of the 
traditional collaborative filtering algorithm. In conclusion, 
the ANN algorithm was effective. 
Although some results have been achieved in the 
study of intelligent course recommendations, there are still 
some shortcomings. For example, the data used in the 
experiments were offline data, which did consider the real-
time nature of the recommendation; the selection of data 
and the determination of weights were also subjective. 
These problems need to be addressed in future work. 
5 Conclusion 
This paper analyzed the intelligent course 
recommendation for college students’ innovation and 
entrepreneurship education based on neural networks and 
compared the performance of the traditional collaborative 
filtering algorithm and the ANN algorithm on course 
recommendation. The experiments on the same data set 
found that the ANN algorithm not only could train and test 
data faster but also had better results in recommendation; 
the HR and NDCG values of the ANN algorithm were 
significantly higher than those of the traditional 
collaborative filtering algorithm, which verified the 
reliability of the method. The ANN algorithm can be 
further promoted and applied in practice. 
References 
[1] Upendran D, Chatterjee S, Sindhumol S, Bijlani K 
(2016). Application of Predictive Analytics in 
Intelligent Course Recommendation. Procedia 
Computer Science, 93, pp. 917-923.  
https://doi.org/10.1016/j.procs.2016.07.267 
[2] Zhu YF, Lu H, Qiu P, Shi KZ, Chambua J, Niu ZD 
(2020). Heterogeneous teaching evaluation network 
based offline course recommendation with graph 
learning and tensor factorization - ScienceDirect. 
Neurocomputing, 415, pp. 84-95.  
https://doi.org/10.1016/j.neucom.2020.07.064 
[3] Kumar T S, Pandey S (2015). Customization of 
Recommendation System Using Collaborative 
Filtering Algorithm on Cloud Using Mahout. 
Advances in Intelligent Systems and Computing, 321, 
pp. 39-43.  
https://doi.org/10.15623/ijret.2014.0319008 
[4] Zhang Z J, Liu H (2015). Research on context-
awareness mobile SNS recommendation algorithm. 
Pattern Recognition and Artificial Intelligence, 28, 
pp. 404-410.  
https://doi.org/10.16451/j.cnki.issn1003-
6059.201505003 
[5] Liu X (2017). An improved clustering-based 
collaborative filtering recommendation algorithm. 
Cluster Computing, 20, pp. 1281-1288. 
[6] He Y, Tan J (2015). Study on SINA micro-blog 
personalized recommendation based on semantic 
network. Expert Systems with Applications, 42, pp. 
4797-4804. 
https://doi.org/10.1016/j.eswa.2015.01.045 
[7] Tan F, Li L, Zhang ZY, Guo YL (2016). A multi-
attribute probabilistic matrix factorization model for 
personalized recommendation. Pattern Analysis & 
Applications, 19, pp. 857-866.  
https://doi.org/10.1007/978-3-319-21042-1_57 
[8] Zhang H, Huang T, Lv Z, Liu S, Zhou Z (2018). 
MCRS: A course recommendation system for 
MOOCs. Multimedia Tools & Applications, 77, pp. 
7051-7069.  
https://doi.org/10.1007/s11042-017-4620-2 
[9] Li C, He K (2017). CBMR: An optimized 
MapReduce for item-based collaborative filtering 
recommendation algorithm with empirical analysis. 
Concurrency and Computation: Practice and 
Experience, 29, pp. e4092.  
https://doi.org/10.1002/cpe.4092 
[10] Tan YL, Zhao HJ, Wang YR, Qiu M (2018). 
Probability matrix decomposition based 
collaborative filtering recommendation algorithm. 
Informatica, 42, 265-271. 
[11] Koprinkovahristova P, Mladenov V, Kasabov N K 
(2015). Artificial Neural Networks. European 
Urology, 40, pp. 245. https://doi.org/10.1002/1097-

100 Informatica 46 (2022) 95–100  J. Zou 
0142(20010415)91:8%2B<1615::AID-
CNCR1175>3.0.CO;2-L 
[12] Jain A K, Mao J, Mohiuddin K M (2015). Artificial 
neural networks: a tutorial. Computer, 29, pp. 31-44. 
https://doi.org/10.1109/2.485891 
[13] Dalila C, Badis EAO, Saddek B, Naït-ali A (2020). 
Feature level fusion of face and voice biometrics 
systems using artificial neural network for personal 
recognition. Informatica, 44, 85-96.  
https://doi.org/10.31449/inf.v44i1.2596 
[14] Jayashree R, Christy A (2015). Improving the 
Enhanced Recommended System Using Bayesian 
Approximation Method and Normalized Discounted 
Cumulative Gain. Procedia Computer Science, 50, 
pp. 216-222.  
https://doi.org/10.1016/j.procs.2015.04.057 
[15] Huy HNL, Minh HH, Van TN, Van HN (2021). 
Keyphrase extraction model: a new design and 
application on tourism information. Informatica, 45, 
563-569. 
[16] Balasingam U, Palaniswamy G (2019). Feature 
augmentation based hybrid collaborative filtering 
using tree boosted ensemble. Informatica, 43, 477-
483. https://doi.org/10.31449/inf.v43i4.2141 
[17] Alhamid M F, Rawashdeh M, Dong H, Hossain MA, 
Saddik AE (2017). Exploring Latent Preferences for 
Context-Aware Personalized Recommendation 
Systems. IEEE Transactions on Human-Machine 
Systems, 46, pp. 1-9.  
https://doi.org/10.1109/THMS.2015.2509965 
[18] Zhang J, Lin Y, Lin M, Liu J (2016). An effective 
collaborative filtering algorithm based on user 
preference clustering. Applied Intelligence, 45, pp. 
230-240. https://doi.org/10.1007/s10489-015-0756-
9 
[19] Gao H, Qiu Z, Liu Z, Huang L, San Y (2016). 
Research and Practice on College Students' 
Innovation and Entrepreneurship Education. 
International Conference of Pioneering Computer 
Scientists, Engineers and Educators, pp. 14-15. 
https://doi.org/10.1007/978-981-10-2098-8_6 
[20] Zhou Z (2016). Research on Model Construction of 
Innovation and Entrepreneurship Education in 
Domestic Colleges. Creative Education, 7, pp. 655-
659. https://doi.org/10.4236/ce.2016.74068