基于拓扑结构与相似度信息融合的教育知识图谱节点重要性评估模型OA

Educational knowledge graph is an important tool for representing relationships between knowledge and concept in the field of education.Understanding and evaluating the importance of node in educational knowledge graph are crucial for tasks like educational resource management and learning path recommendation.However,traditional node importance evaluation methods often treat all nodes equally,considering only certain topological structures,and fail to capture the comprehensive characteristics of node.To address this,we proposed TSFM driven by key node.TSFM evaluated node importance by integrating topological structure with semantic and structural similarity information driven by key node.Specifically,TSFM utilized graph neural networks to embed nodes in knowledge graph,and optimized the embedding representation by considering topological and semantic similarity between nodes.Experimental results demonstrate that TSFM outperforms traditional algorithms and recent graph neural network models across multiple evaluation metrics.