Automatic Curriculum Cohesion Analysis Based on Knowledge Graphs
摘要
Creating a well-structured and cohesive curriculum is essential for providing students with a meaningful and organized educational experience. However, the complexity and scale of curricula and the need for regular updates make developing a curriculum for a single major a highly challenging task. In this paper, we aim to construct a comprehensive tool designed to automatically analyze existing curricula and identify areas of incoherence. The presented approach utilizes unsupervised learning techniques and large language models (LLMs) to extract the main concepts covered by courses and their prerequisites directly from course syllabi. The extracted entities are then linked to Wikidata entities, ensuring both their correctness and uniqueness. By constructing a knowledge graph based on the relationships between courses and main educational concepts, the cohesion of the curriculum can be systematically verified. The verification process involves checking if all cohesion logic predicates are satisfied, thereby ensuring that the curriculum is both comprehensive and logically structured. The results demonstrate the efficacy of our approach in identifying inconsistencies within educational curricula, thereby aiding in the development of more cohesive and well-organized educational programs.