Dynamic Educational Knowledge Graph Model via Information Entropy for Knowledge Building

Knowledge building is the production and continual improvement of ideas of value to a community, which attaches importance to conceptual engagement and contribution. However, knowledge building community will accumulate large and complex semi-structured educational data over time. It is not conducive to the continuation of in-depth knowledge building activities. To overcome these issues, we propose a dynamic educational knowledge graph with information entropy (IE-DEKG) model for knowledge building community. The model can construct dynamic knowledge graphs that contain instructional concepts and educational relations for learners. Specifically, it adopts the mutual information and adjacent information entropy to detect new terminologies on pedagogical data, and then the topic modeling algorithm is utilized to extract instructional concepts. Moreover, the model employs association rule mining to identify the prerequisite relations and uses pattern matching to obtain the inclusion relations. For the sake of satisfying the needs of educational applications and services, we design and implement the dynamic educational knowledge graph system. Experimental results demonstrate that the proposed IE-DEKG method outperforms the state-of-the-art methods.