Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

Opening Knowledge Graph Model Building of Artificial

Intelligence Curriculum
https://doi.org/10.3991/ijet.v17i14.32613

Hongwei Yue1, Hanhui Lin2,3(), Yingying Jin4, Hui Zhang1, Ken Cai5
1 Guangdong University of Education, Guangzhou, China
2 Guangdong University of Finance and Economics, Guangzhou, China
3 South China Normal University, Guangzhou, China
4 Guangzhou Panyu Polytechnic, Guangzhou, China
5 Zhongkai University of Agriculture and Engineering, Guangzhou, China

[email protected]

Abstract—The knowledge points setting of artificial intelligence curriculum

has shortcomings in connection between theory and practices. To overcome the
problem, this study designs an open knowledge point design model based on
knowledge graph. Fist, to promote the construction of the knowledge graph
(KG) of curriculums, associated teaching research was analyzed visually. Then
the order and hierarchical structure of the knowledge points were defined, and
the ontology structure of curriculum knowledge and the relationship between
knowledge points and posts were designed as well. Moreover, an overall logic
structure for the construction of the open KG of curriculums was proposed. Re-
sults demonstrated that high attention should be paid to the construction and
concern of teaching teams for artificial intelligence algorithms and the KG of
curriculum construction. Additionally, the opening model can strengthen the
openness of the KG of curriculums to reinforce the close connections between
classroom knowledge and practices. Research conclusions are conducive to un-
derstand the existing problems in the KG of curriculums and provide beneficial
references to the integration of information technology and education.

Keywords—artificial intelligence, KG of curriculums, university education,

educational informationization

1 Introduction

With the development of cloud computing and big data, online education has
changed the traditional teaching mode. Various mainstream education platforms
worldwide have launched abundant learning resources in different forms, such as text,
audio, video. Online course learning has become one of the most common learning
modes. Students, life-long learners, and adult education learners can all use fragment-
ed time and acquire considerable learning resources according to their learning de-
mands.
At present, building educational informationization based on the new generation of
information technology has become the essential trend. Among them, the knowledge

64 http://www.i-jet.org
Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

graph (KG) is attracting increasing attention [1]. According to application fields, KG

usually is divided into two types: the general-purpose knowledge graph (GKG) and
the domain-specific knowledge graph (DKG). For the GKG, many scholars have
carried out deep studies on key technologies related to the construction and learning
of KG, knowledge evolution, and reasoning. The multi-element, multisubject, and
multidata-source KG construction has achieved rapid development [2].
In university education, artificial intelligence as an important major is characteris-
tic of diversified information, diversified learning modes, and diversified curriculums,
which also bring many problems. (1) With respect to the training mode, students ac-
cept different programming requirements for different curriculums. The construction
of most teaching teams lacks a uniform management mode, and students are tired of
the follow-up use of programming platforms and tools. (2) The artificial intelligence
curriculum involves a lot of teaching content, and complicated relations exist among
knowledge points. Textbooks contents emphasize the introduction of basic concepts,
classical algorithms, and applications, which are boring and disconnected from practi-
cal applications at present. All of these problems are barriers against the development
of university education even though the education resource development of universi-
ties and colleges are of substantial concern among education researchers and users.
For both traditional education and new education based on information technology,
education itself is a data-intensive industry. The essence of KG is to connect things
and concepts through interactions [3,4]. This characteristic implies that the curricu-
lums of artificial intelligence in university education can reconstruct connections
among knowledge from abundant disordered information resources based on KG and
a knowledge system of different curriculums can be established effectively.
Therefore, analysis and research on curriculum knowledge of artificial intelligence
and building of KG for the curriculums are significant to the knowledge acquisition
and curriculum learning of learners. Notably, the KG of curriculums in the present
study belongs to DKG. The follow-up state of the art is carried out on the basis of the
KG of curriculums.

2 State of the art

KG was first used for searching. In 2012, Google launched the KG project and an-
nounced the construction of a next-generation intelligent search engine based on KG
[5]. KG has promoted the development of a semantic network since it was proposed
in 2012. Many giant scientific and technological companies have applied KG to sup-
port the development of intelligent applications [6]. With respect to education, many
scholars have studied how to use KG in classroom teaching, online teaching, and
higher education.
Most academic studies on the KG of curriculums focus on the following aspects.
First, research hotspots and leading edges have been analyzed using KG. Shao et al.
[7] discussed and analyzed research hotspots in the field of recommendation systems
and corresponding characteristics. Meanwhile, Mu et al. [8] analyzed research devel-
opment on mechanical engineering practice teaching and discussed current research

iJET ‒ Vol. 17, No. 14, 2022 65

Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

hotspots as well as existing problems. In addition, Liu et al. [9] helped postgraduates
to understand quickly the professional subjects in which they were interested on the
basis of KG, thus improving the scientific research level of postgraduates.
Second, network relations among subject knowledge points were established using
KG tools. Seo et al. [10] proposed how to provide learning paths for learners. Fur-
thermore, Karas et al. [11] proposed a method to help students to form the knowledge
network model and analysis of KG. In another study, Alshalabi et al. [12] proposed an
adaptive learning path algorithm by analyzing the structure of domain-specific
knowledge to improve the academic performance of students. Moreover, Lv et al. [13]
solved the lack of logic knowledge relations on the semantic level using the nursing
KG of the curriculum model. Chen et al. [14] proposed the knowledge mapping sys-
tem in the education field to extract the relationship between teaching concepts and
implicit education from heterogeneous data sources.
Third, some scholars have explored the recommendation of education resources on
the basis of KG. In particular, KG-based recommendation system focus on education
resources or relevant academic contents, which can be integrated into the learning
process in [15,16].
Existing studies on network relations among knowledge points in the education
field mainly focus on KG among different subjects. This type of KG generally has
problems on single nodes, single relations, and single attributes, as well as the coarse
division granularity of knowledge points. Nevertheless, the KG of curriculums is
more important to students in practical teaching processes. Students can have supple-
mental learning by searching for knowledge points or basic knowledge points similar
to the current knowledge points according to KG. They can also search for the next
knowledge point for learning according to KG once a knowledge point has been mas-
tered. In other words, the existing DKG models generally have insufficient generali-
zation ability and poor ability. Some scholars have also studied the KG of specific
curriculums thoroughly and developed many KG cases of curriculums [17,18].
With the scientific technological development, theories and technologies of KG are
becoming increasingly perfect and mature, and its application ranges are expanding
gradually [19]. To master the application status of the KG of curriculums in teaching
accurately, studies about the KG of curriculums have been summarized comprehen-
sively and systematically through bibliometrics and information visualization. The
studies are expected to show research hotspots and trends intuitively. Subsequently,
the design framework of the open KG of curriculums is proposed. This framework is
of importance to teaching design based on KG and training application-oriented inno-
vative talents in China.
The reminder of this study is organized as follows. Section 3 analyzes and discuss-
es studies concerning the KG of curriculums. Then, Section 4 proposes the design
framework of the open KG of curriculums. Finally, Section 5 summarizes the conclu-
sions.

66 http://www.i-jet.org
Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

3 Methodology

3.1 Data source

Data from January 2013 to December 2021 were collected from the Chinese Na-
tional Knowledge Infrastructure (CNKI). Studies in the journal were searched using
the keyword “knowledge graph of curriculums”. Irrelevant studies were excluded by
reading the abstracts manually, and a total of 168 studies were searched. Finally, the
data were output in EndNote using the format of CNKI and were stored as preparation
for the next studies. Subsequently, quantitative statistics and analyses on the number
of studies published, keywords, and other characteristics were carried out using SPSS
Statistics and UCINET analytical tools.

3.2 Comparison and analysis

Time zone and theme distribution of articles. The quantitative distribution of
annual literature studies can reflect the state of the art and the overall outcomes of the
field to some extent. To make data more intuitive, a statistical analysis on relevant
studies published from January 2013 to December 2021 was carried out, through
which the tendency chart of the quantity of the annual academic literature was ob-
tained (Figure 1). In Figure 1, the red and blue tendency lines represent the annual
distributions gained by using “knowledge graph of curriculums” and “knowledge
graph” as keywords, respectively. Evidently, KG has extensive applications in other
fields, but it is relatively rarely used in classroom education. Studies on the KG of
curriculums in universities and colleges in China started in 2013, and an extremely
small quantity of studies have been published in this field. However, the trend of
curves shows that researchers are becoming increasingly interested in the applications
of KG in teaching fields, and the relevant literature is growing quickly.
3000 60

2000 40

1000 20

0 0
2013 2014 2015 2016 2017 2018 2019 2020 2021

Fig. 1. Annual distribution of academic literatures related to KG

Subsequently, theme distributions related to the KG of curriculums were analyzed

in this study. The results are shown in Figure 2. Clearly, themes that are mostly relat-
ed to the research contents of this study can be summarized as knowledge graph anal-
ysis, visual analysis, knowledge graph construction, and curriculum development. All

iJET ‒ Vol. 17, No. 14, 2022 67

Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

of these themes can reflect the state of the art intuitively. Notably, studies related to
curriculum development account for 2.96%. This amount also indicates that although
KG has attracted some attention and achieved some development in the field of edu-
cation, few curriculum-based KG cases are available.

2(1.18%)

64(37.87%)
knowledge graph
……
knowledge graph analysis
research hotspot
……
visual analysis
knowledge graph construction
curriculum development
5(2.96%) co-word matrix
5(2.96%) physical education
8(4.73%) 17(10.06%)
11(6.51%)

Fig. 2. Research theme distribution of literature related to KG of curriculums

Literature publishing conditions of authors. The co-occurrence network graph

of authors can reflect the cooperative relations of authors in this area of research.
Figure 3 demonstrates that teaching researcher groups are scattered and the number of
published studies is not concentrated. The cooperative relation among different re-
search teams is relatively weak. The Table 1 shows that F. Sun from Wenzhou Uni-
versity published the most articles (four articles) in the education field, while most
authors published less than two articles. During the literature review, most research
teams were composed of members from the same institution as the author, and the
cooperation among institutions is relatively weak. On one hand, the above situations
demonstrate that many scholars engage in the KG-related education field and take the
initiative to explore teaching reforms. On the other hand, the circumstances also prove
that no core author group has been formed. Moreover, most authors lack continuous
concern of the KG-related education field.

Table 1. Relevant information of authors with high publications

Sorting Author Num. (papers) Subordinate institution
1 F. Sun 4 Wenzhou Univ.
2 L. Pan 3 Wenzhou Univ.
3 P. Cheng 2 Chongqing Univ. of Tech
3 H. Peng 2 Wuhan Univ. of Tech
3 G. Zhang 2 Wuhan Univ. of Tech

68 http://www.i-jet.org
Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

D. Sun
D. Shao
Y. Xie H. Hu J. Luo
J. Yuan
H. Zhang F. Sun
X. Hou
L. Pan Y. Bai
Q. Chen Z. Zhou
Z. Liu S. Gao B. Hao
Y. Li
Y. Fan H. Peng
T. Ma
K. Hu
X. Zhao S. Han
Z. Qiang
H. Li L. Wu R. Ni Y. Wang G. Zhang
X. Ma
Y. Wu Z. Shi
J. Yang
Q. Liu J. Cheng
L. Jiang
S. Yuan X. Li
B. Li

Fig. 3. Co-occurrence network graph of authors

Research hotspot analysis based on keywords. Research hotspots refer to re-

search problems or subjects discussed by a group of papers that have internal connec-
tions and a high quantity within a period. Keyword refers to the high extraction and
summary of the author to core contents. It reflects the core contents and important
information of a study. Hence, keywords that occur frequently can be used to deter-
mine hotspots in a research field.
The keyword co-occurrence network can reflect the research hotspot and develop-
ment trend of a field. In this study, the keywords used in studies were analyzed using
SPSS Statistics. Overall, 496 keywords were collected. Among them, high-frequency
keywords were screened and used to construct the co-occurrence matrix of keywords.
This co-occurrence matrix can reveal internal connections among keywords. The
formula of transformation into the similar matrix is:

N ik
Sik = (1)
( N i * N k )1/2
Where N ik is the co-occurrence frequency of two keywords, is the occurrence fre-
quency of the keyword i , and N k is the occurrence frequency of the keyword k .
According to the formula, the co-occurrence matrix can be constructed according to
this formula. If the number is closer to 1, the two keywords are more similar; other-
wise, the two keywords are less similar.
To discover important information in the data and explore internal connections of
hotspots in the KG-related education field, the extracted co-occurrence matrix was
input into the UCINET software, and the co-occurrence graph of keywords was plot-

iJET ‒ Vol. 17, No. 14, 2022 69

Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

ted (Figure 4). Figure 4 demonstrates that keywords, such as “flipped class-
room”,“visual analysis”,“personalized learning”,“learning path”,“online course” and
“MOOC” were all surrounding “KG”.Moreover, they had a short relative distance
with “KG”. This observation proves that the nodes where these keywords were locat-
ed were closely related. In contrast, keywords like “core literacy”, “artificial intelli-
gence” and “curriculum reform” were at edges of the co-occurrence graph, thus indi-
cating that they might be research hotspots in future.

course knowledge graph

online learning entity identification
deep learning
recommendation algorithm
personalized recommendation adaptive learning
learning path artificial intelligence
knowledge graph intelligent education
MOOC
online course personalized learning

learning resource recommendation

visual analysis
graph databases
big data flipped classroom
co-word analysis

knowledge point

curriculum system core literacy

Fig. 4. Co-occurrence network graph of keywords with high frequency

Clustering analysis of high-frequency keywords. For the convenience of exhibi-

tion, the top 18 high-frequency keywords were clustered in the co-occurrence matrix.
After the type (pedigree), method (intergroup connection), and measurement standard
interval (square Euclidean distance) of the graph were set, the keywords clustering
graph is shown in Figure 5. The figure shows that these two keywords were divided
into two clusters. According to the meaning of keywords in each cluster, the tag of
each cluster was defined as “Cluster#1:individualized education” and “Clus-
ter#2:curriculum development”. Cluster#1 mainly reflected the applications of various
artificial intelligence algorithms in personalized learning, path recommendation, and
logic reasoning. Cluster#2 mainly reflected how to construct the KG of curriculums
by using existing curriculum standards, disciplinary knowledge, and teaching re-
sources. Both clusters reflected two research directions of KG in China.

70 http://www.i-jet.org
Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

0 5 10 15 20 25
r #1 learning path 7
alized recommendation 13
ourse knowledge graph 9
personalized learning 5
intelligent education 15
artificial intelligence 6
adaptive learning 12
curriculum reform 2
core literacy 11
r #2 MOOC 1
big data 10
flipped classroom 3
online course 8
co-word analysis 18
visual analysis 4
knowledge point 16
graph databases 14
curriculum system 17

Fig. 5. Clustering results of keywords

According to the previous statistical analysis, some problems are summarized as

follows. (1) KG construction of curriculums is still in an exploration stage because the
KG of curriculums belongs to DKG. On one hand, the construction of DKG requires
professional field knowledge. On the other hand, the data source in the professional
field has a small size, and it cannot meet the needs of the data size. (2) The KG con-
struction of curriculums is closed. Such closeness not only means that few teaching
teams are cooperating with other institutions, but it also entails teaching knowledge
points and delayed market demands in the setting. Upon reading the literature, the
author also found that the KG construction of curriculums has a low quality at pre-
sent. For example, the teaching objective, teaching design, and teaching mode are
limited within institutions to which teachers are affiliated, professional student groups
they face during daily teaching, and the accumulated teaching experiences of teachers
or teaching teams.
In this study, the first problem that the KG of curriculums faces will be solved in
time. In the following text, how to design an open KG of curriculums will be dis-
cussed.

4 Result analysis of open KG design of curriculums

How to construct an open KG is discussed through a case study of artificial intelli-

gence in university education. Artificial intelligence has been selected as an example
mainly because it is an integrated discipline with highly disciplinary crossing and
integration involving machine learning, pattern recognition, data mining, and neural
network. Moreover, artificial intelligence has been applied extensively [20]. Although
many universities and colleges have carried out a teaching reform for artificial intelli-

iJET ‒ Vol. 17, No. 14, 2022 71

Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

gence in recent years, they still have shortcomings in connecting theory with practice
[21].
To solve these problems, the training scheme and curriculum content have been
formulated together with enterprise engineers according to the post requirements, thus
finishing the KG design. Teachers and engineers cooperate in setting the overall logic
structure for an open KG (Figure 6). (1) Clearly, teachers use teaching resources,
CNKI, and Baidu as major data sources. After the semantic analysis of the resource
description library based on semantic analysis tools, keywords were extracted, and the
semantic description model and KG based on structures were constructed. (2) Mean-
while, enterprise engineers mainly use the occupational ability of artificial intelli-
gence posts as the major data source. Combining with knowledge points setting by
curriculum teachers, network key nodes were built through semantic tools for seman-
tic description, and an industrial KG based on professional skills was built.
As KG is a structured semantic knowledge library, the ontological structural design
of the KG of curriculums is the key link. In the following text, the curriculum re-
source ontological model design was taken as an example to elaborate how teachers
and engineers construct KG (Figure 7). The KG of curriculums is an important refer-
ence to select the learning content and recommend a learning path. According to the
requirements of enterprise engineers for occupational ability, the knowledge points
that the curriculum required were analyzed. To assure that learners could choose
learning contents within the controllable range, the KG of curriculums used chapter,
section, and knowledge points as knowledge units and involved three semantic rela-
tions: “inclusion relation”, “fore-and-aft relation” and “parallel relation” [17]. In this
study, the curriculum knowledge structural model defines the order and hierarchical
relations of knowledge points [18].
Definition 1: Knowledge point ( Ke ) model: This knowledge point cannot be cut
into other knowledge points, that is, the lowest semantic level.
Definition 2: Knowledge chain ( Kc ) model: It is the superior layer of the
knowledge point. The knowledge chain is the chain structure formed by knowledge
points according to different logic relations (Eq.(2)):

< ( K ei , K ek ), R >
Kc = (2)

where Kei and Kek are different knowledge points. R refers to the relationship among
knowledge points, which are divided into order relations and hierarchical relationship.
Definition 3: Knowledge unit ( Ku ): It is the superior layer of the knowledge chain,
and it is formed by knowledge chains (Eq. (3)).

K u =(K ci |i = 1, 2,..., N ) (3)

Definition 4: Course knowledge entity model (CKEM, Kckem ): It is the superior

layer of a knowledge unit, and it is formed by different knowledge units (Eq. (4)).

K ckem =(K ui |i = 1, 2,..., N ) (4)

72 http://www.i-jet.org
Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

In this study, learning contents in the curriculum form the CKEM according to
“knowledge points, knowledge chains, knowledge units.” When the ontology model
of the curriculum is determined, the hierarchical system of the curriculum ontology
can be acquired using the Protégé tool to define the conceptual attributes and data
attributes. The final model (e.g., type, example, relation and attributes) is stored in
RDF format, and it is input into the graph database Neo4j. As this study focuses on
the state of the art and proposes solutions to existing problems in the KG of curricu-
lums, it has not introduced methods to establish KG for a curriculum.

Course 1:Python KP 1.1 KP 1.n

language programming LG 1.1 ... LG 1.n
Course 2:Digital image KP 2.1 KP 2.n
processing LG 2.1 ... LG 2.n
Course 3:Machine KP 3.1 KP 3.n
learning LG 3.1 ... LG 3.n

Course 4:Deep learning KP 4.1 KP 4.n

LG 4.1 ... LG 4.n

Job position 1: Computer PA 1.1

vision ... PA 1.n
Job position 2 : Machine PA 2.1
learning ... PA 2.n
Job position 3 : Natural PA 3.1
Language Processing ... PA 3.n
Semantic Job position 4 : Data PA 4.1
Mining ... PA 4.n
Engineer Database analysis tool

PA:professional ability
Kn
LG:learning goal v
KP:knowledge point

Fig. 6. Overall logic structure of open KG

iJET ‒ Vol. 17, No. 14, 2022 73

Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

C:chapter C1 C2 …… Cn

S:section S1.1 S1.2 S2.1 S2.2 …… Sn.1 Sn.2

K:knowledge K K K K K …… K K K

inclusion relation K K K K
fore-and-aft relation
parallel relation Job position 1 Job position 2 …… Job position n

Fig. 7. Structure of curriculum knowledge and relations between knowledge points and jobs

5 Discussion and conclusion

Connecting relations among knowledge points can be reconstructed through the

KG of artificial intelligence curriculums, thus forming a professional networked struc-
tural system of the subject to support flexible, accurate “teaching” and lifelong per-
sonalized “learning” effectively. To understand the state of the art of the KG of cur-
riculums, improve the innovation and academic and application values of research
contents, and avoid repeated studies, a statistical analysis of studies concerning the
KG of curriculums was carried out, and an open KG design scheme was proposed.
The following conclusions could be drawn:
1. Few constructions of KG for a specific curriculum have been established, and KG
is still in its initial stage. This state can be explained through the following two as-
pects. Curriculum knowledge has very complicated types and structures; some cur-
riculum knowledge is even difficult to structuralize, and the workload is very high.
Moreover, the diversified changes of some subjects increase difficulties in forming
gradient teaching teams, thus resulting in the failure of collaborative teaching. This
study believes that this problem can be improved through the following aspects.
First, the construction of teaching research teams must be strengthened; leaders,
managers, theory teachers, and professional experimental technicians of various
disciplines must also be encouraged to form a practice teaching team with reasona-
ble age and knowledge structures, form academic gradient teams with continuous
concern for practice teaching research, assure ordered progress in teaching reforms,
and publish papers on research results continuously. Second, the attention of uni-
versities, especially top-class universities, on studies of practice teaching should be
improved, and a good atmosphere for practice teaching should be created.
2. The KG of curriculums has two research directions. One is to explore the applica-
tions of artificial intelligence algorithms in KG construction. The other is to study
the specific implementation of the KG construction of curriculums. Under these

74 http://www.i-jet.org
Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

two directions, researchers propose the application demonstrations of education-

related KG in various scenes, such as intelligent teaching and intelligent learning.
3. The design of an open KG of artificial intelligence curriculums in university educa-
tion is conducive to make knowledge point setting conform to enterprise needs.
Through cooperation between teachers and engineers, boring classroom contents
are connected with the requirements of a specific post, which is beneficial to im-
prove the practice ability of students and avoid disconnection with practical appli-
cations. This cooperation is also conducive to the future career plans of learners.
The current research on leading edges and tendency related to the KG of curricu-
lums is analyzed in this study. Results provide a reliable perspective and reference to
education researchers and practitioners. However, this study is based on the existing
literature on the CNKI database. Future studies can further analyze foreign studies
and compared them with the relevant Chinese literature to provide references for
further localized teaching practices. In this way, research results can serve the educa-
tion field better. Moreover, the future development of intelligent education involves
knowledge expression learning, knowledge acquisition, disciplinary knowledge rea-
soning, and computation. Deep studies on the basic theories and methods of large-
scale education KG are some research directions in future.

6 Acknowledgment

This study was supported by the Education and Scientific Research Project of
Guangdong Province(NO.2020GXJK482),the Natural Science Foundation of Guang-
dong Province(NO.2020A1515110458) ,the Education and Teaching Reform Project
of Guangzhou Panyu Polytechnic(No.2022JG09),the Collaborative Education Project
of Higher Education of Education Ministry(NO.201902118021),the Teaching Quality
and Reform Project of Guangdong University of Education(NO.2019xnfzsy01),the
Collaborative Education Project of Guangdong Educational Department
(NO.1112626696093306880),the Scientific Research Project of Guangzhou Panyu
Polytechnic(No.2022KJ02),the Jiangmen Basic and Applied Basic Research Founda-
tion(No.2021030102570004880),the Haizhu District Science and Technology Plan-
ning Project of Guangzhou(NO.202271802).

7 References
[1] Rizun, M. (2019). Knowledge graph application in education: a literature review. Acta
Universitatis Lodziensis, 3(342): 7-19. http://doi.org/10.18778/0208-6018.342.01
[2] Velampalli, S., Jonnalagedda, M. V. (2017). Graph based knowledge discovery using
MapReduce and SUBDUE algorithm. Data & Knowledge Engineering, 111: 103-113.
https://doi.org/10.1016/j.datak.2017.08.001
[3] Paulheim, H. (2017). Knowledge graph refinement: A survey of approaches and evaluation
methods. Semantic Web, 8(3): 489-508. http://doi.org/10.3233/SW-160218
[4] Nguyen, H. L., Vu, D. T., Jung, J. J. (2020). Knowledge graph fusion for smart systems: A
survey. Information Fusion, 61: 56-70. http://doi.org/10.1016/j.inffus.2020.03.014

iJET ‒ Vol. 17, No. 14, 2022 75

Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

[5] Ernst, P., Siu, A., Weikum, G. (2015). Knowlife: a versatile approach for constructing a
large knowledge graph for biomedical sciences. BMC Bioinformatics,16(1): 1-13.
https://doi.org/10.1186/s12859-015-0549-5
[6] Noy, N., Gao, Y., Jain, A., Narayanan, A., Patterson, A., Taylor, J. (2019). Industry-scale
knowledge graphs: lessons and challenges. Communications of the ACM, 62(8): 36-43.
http://doi.org/10.1145/3331166
[7] Shao, B., Li, X., Bian, G. (2021). A survey of research hotspots and frontier trends of rec-
ommendation systems from the perspective of knowledge graph. Expert Systems with Ap-
plications, 165: 113764. http://doi.org/10.1016/j.eswa.2020.113764
[8] Mu, G., Zhang, G., Zhang, H., Pan, L., Zhang, Q., Li, X., Ding, X., Gao, W. (2021).
Knowledge mapping analysis of mechanical engineering practical teaching research in
China based on CiteSpace. Higher Agricultural Education, 3: 83-88. http://doi.org/10.1383
9/j.cnki.hae.2021.3.013
[9] Liu, Y., Zhen, P., Li, A. (2016). Application of scientific knowledge map in graduate stu-
dents’ embedded information literacy education. Journal of Library and Information Sci-
ence, 1(10): 145-148. http://doi.org/10.3969/j.issn.1005-6033.2016.10.037
[10] Seo, S., Oh, B., Lee, K. H. (2020). Reliable knowledge graph path representation learning.
IEEE Access, 8: 32816-32825. http://doi.org/10.1109/ACCESS.2020.2973923
[11] Karas, S., Konev, A. (2017). Knowledge engineering as a component of the curriculum for
medical cybernetists. Studies in Health Technology and Informatics, 236: 241-247.
http://doi.org/10.3233/978-1-61499-759-7-241
[12] Alshalabi, I. A., Hamada, S. E., Elleithy, K. M., Badara, J. A., & Moslehpour, S. (2018).
Automated Adaptive Mobile Learning System using Shortest Path Algorithm and Learning
Style. International Journal of Interactive Mobile Technologies, 12(5): 4-27. http://doi.org/
10.3991/ijim.v12i5.8186
[13] Lv, Q., Zhang, Y., Li, Y., Yu, Y. (2022). Research on a Health Care Personnel Training
Model Based on Multilayered Knowledge Mapping for the Integration of Nursing Courses
and Examinations. Journal of Healthcare Engineering. http://doi.org/10.1155/2022/382641
3
[14] Chen, P., Lu, Y., Zheng, V. W., Chen, X., & Yang, B. (2018). Knowedu: A system to con-
struct knowledge graph for education. IEEE Access, 6: 31553-31563. http://doi.org/10.110
9/ACCESS.2018.2839607
[15] Huang, Y., Zhu, J. (2021). A Personalized English Learning Material Recommendation
System Based on Knowledge Graph. International Journal of Emerging Technologies in
Learning, 16(11): 160-173. http://doi.org/10.3991/ijet.v16i11.23317
[16] Li, X., Chen, Y., Pettit, B., Rijke, M. D. (2019). Personalised reranking of paper recom-
mendations using paper content and user behavior. ACM Transactions on Information Sys-
tems, 37(3): 1-23. http://doi.org/10.1145/3312528
[17] Cao, F. (2021). Research and Design on the Learning Resources Recommendation System
Based on Knowledge Graph. Modern Information Technology, 5(04):126-128+132.
http://doi.org/10.19850/j.cnki.2096-4706.2021.04.032
[18] Yuan, M., Qiu, T., Hu, C. (2019).Fine-Grained Course Knowledge Meta-Organization
Model and Knowledge Graph Implementation.Journal of Jilin University(Information Sci-
ence Edition), 37(05): 526-532. http://doi.org/10.19850/j.cnki.2096-4706.2021.04.032
[19] Yoo, K. (2021). Knowledge graph-based knowledge map for efficient expression and in-
ference of associated knowledge. Journal of Intelligence and Information Systems, 27(4):
49-71. http://doi.org/10.13088/jiis.2021.27.4.049

76 http://www.i-jet.org
Paper—Opening Knowledge Graph Model Building of Artificial Intelligence Curriculum

[20] Kao, C. H. (2021). Enriching Undergraduate Mathematics Curriculum with Computer Sci-
ence Courses. International Journal of Engineering Pedagogy, 11(5):37-53. https://doi.org/
10.3991/ijep.v11i5.21701
[21] Long, X., Dai, A. (2016). Artificial intelligence+education:a new opportunity for the
reformation of talent training in Chinese universities. University Education Science,4:107-
113. https://doi.org/10.3969/j.issn.1672-0717.2019.04.016

8 Authors

Hongwei Yue is an associate professor in the School of Physics and Information

Engineering, Guangdong University of Education. His research interests include im-
age processing and technical education.
Hanhui Lin (Corresponding author) is a senior engineer in the Center of Network
Information and Educational Technology, Guangdong University of Finance and
Economics and in School of Information Technology in Education, South China
Normal University. His research interests include teachers training and educational
technology.
Yingying Jin is an associate professor in the School of Common Course, Guang-
zhou Panyu Polytechnic. Her research interests include General topology and multi-
media courseware.
Hui Zhang is a lecturer in the School of Physics and Information Engineering,
Guangdong University of Education. His research interests include embedded system
development and application.
Ken Cai is with the College of Automation, Zhongkai University of Agriculture
and Engineering. His research interests include image processing and technical educa-
tion.

Article submitted 2022-04-21. Resubmitted 2022-05-24. Final acceptance 2022-05-25. Final version
published as submitted by the authors.

iJET ‒ Vol. 17, No. 14, 2022 77