Graph neural network-driven knowledge graph construction and application for Chinese instruction in higher vocational education

As a structured knowledge representation method, Knowledge Graph (KG) can intuitively show the relevance of knowledge, but the traditional construction method has some problems such as low efficiency and poor expansibility. In this paper, a construction method of Higher Vocational Colleges (HVC) Chinese teaching KG based on graph neural network (GNN) is proposed, aiming at breaking through the limitations of traditional methods and realizing automatic association and dynamic updating of knowledge. Through multi-source data acquisition and preprocessing, the bilstm-crf model is used for named entity recognition, and the graph convolution network (GCN) relationship extraction model based on dependency syntax tree is constructed, and the ontology structure including the core concepts of literary works, creation years, genre and so on is constructed. The knowledge reasoning module based on GNN is introduced to realize the dynamic update of KG. Multimodal Graph Attention Network (MGAT) is adopted as the core model architecture, and the model performance is optimized through pre-training and fine-tuning stages. In the teaching application scenario, an intelligent question answering system based on subgraph reasoning and a personalized learning recommendation model are designed. The experimental results show that MGAT model performs well in the tasks of entity identification and relationship extraction, and the dynamic updating efficiency is significantly better than that of traditional methods. The intelligent question answering system and personalized recommendation model perform well in practical applications, effectively improving the teaching effect. This study provides new ideas and methods for HVC Chinese teaching, which has important theoretical and practical significance.