Table 2.

A summary of selected KG construction approaches for curriculum design and planning.

Ref.	KG Specific Purpose	Construction Algorithm(s)	Type of KB	KG Resource(s)	#Entites (e)/#Relations (r)	Evaluation Criteria	Limitation(s)
[62]	Improve curriculum and assist students' learning and understanding.	Water Disciplines Entity-Relationship Joint Extraction (WDERJE) framework	Schema-based	Water conservancy educational big data	#e: NA #r: 180,000	F0.5-score	•Limitations in handling evolving or dynamic knowledge.
[63]	University curriculum design	Manual	Schema-based	Data collected from French students through surveys	#e: 5452 #r: 27	task-based evaluation	•Limited discussion on the construction algorithm. •Poor evaluation techniques,
[64]	integrating social media contents in formal courses	Semantic linking techniques	Hybrid	formal course entities, social media materials	#e: 230#r: N/A	Case study	•The experiment is being conducted within a specific course in the IS domain, which could introduce biases. •The study did not analyze professors' perspectives, which could provide valuable insights. •Some of the social features developed in the tools are not being utilized fully. •The potential for differences in concept labelling by different students in crowdsourced ontology concepts. •Lack of analysis on the implications of dynamically constructed program goals and learning objectives.
[65]	Design a structured representation of curriculum knowledge using multimodal KG.	DeepKE and PaddleOCR	Hybrid	Course materials, teaching videos, and speech content	N/A	Case study	•The DeepKE model's generalization ability might be limited due to the relatively small labelled training data. •The paper doesn't elaborate on the challenges of multimodal fusion, •Lack of quantitative results or analyses of the framework's performance.
[61]	Supporting STEAM learning theme design	BERT and tensor decomposition	Schema-free	Encyclopedia, OpenKG, and national discipline curriculum standards	N/A	P, R, F1, MRR, and Hits@N	•Further research is needed to address the interpretability mechanism of KGs for STEAM interdisciplinary semantic learning.
[66]	Curriculum system improvement for higher education	Various algorithms are used to construct three KGs	Schema-free	syllabuses of courses and external data sources	N/A	Subjective criteria such as positive feedback from teachers and students through surveys.	•The effectiveness of the approach relies on the availability and accuracy of course syllabuses, teacher resumes, and related data. •The approach might require customization for different educational contexts.
[67]	Enhancing MIS course curriculum.	Top-down ontology modeling that maps the ontology model into a KG.	Schema-based	Course material	N/A	Case study	•Limitations in its generalizability to other domains. •Incomplete or inaccurate ontology definitions could lead to limitations in the KG's quality.
[68]	Supporting blended learning	Association mining	Hybrid	Textbooks, curriculum standards, etc.	N/A	Case study	•Lack of adequate description of mechanisms used in constructing and evaluating the KG.
[69]	A visual representation of the curriculum including chapters, sections, and knowledge points.	Vocabulary mining, entity recognition, and knowledge extraction.	Schema-free	High school information technology textbooks	N/A	Case Study	•Potential errors in knowledge extraction, •No discussion on the algorithms used for constructing the graph.
[71]	Optimizing the teaching of internet fraud cases.	Ontology mapping and relational mapping	Schema-based	Internet fraud cases, their characteristics, victims, platforms, fraud processes, technologies, losses, and prevention methods.	N/A	Case study (improvement in students' understanding and awareness of internet fraud)	•Limitations related to the size of the fraud knowledge graph and the potential absence of certain individual cases. •The KGCT model is under development, thus ongoing refinement and improvement are needed.
[72]	Education technology and learning analytics	ontology-based learner modeling techniques	Schema-based	Programming records from an online programming platform	N/A	quasi-experiment, standardized tests, and experimental group	•The study's findings are based on a relatively small sample size of 38 participants, •Lack of technical details of how these graphs are utilized or how the, •The study does not explore the potential impact of KGs on group students or facilitate peer assistance.