Research on the Construction and Appreciation of a Knowledge Graph of Chinese Classical Literature and Culture for International Chinese Language Education

Abstract

As an effective means of gaining a deeper understanding of literary works, this paper proposes the construction of a knowledge graph to assist students in international Chinese language education with the interpretation of classical Chinese literature. The paper provides a comprehensive overview of the knowledge graph construction process and designs a BERT-BiLSTM-CRF model for entity recognition, using labeled sequences as the model’s annotation results. Additionally, a Bs-Spert model is designed for entity relationship extraction, comprising a BERT pre-training module, a clustered search module, a span filtering module, and a relationship classification module. Finally, a knowledge Q&A model is built based on the knowledge graph embeddings to implement the application of the literary graph. The Chinese classical literature knowledge graph includes 12 entity types. The recognition accuracy for different entities is relatively high, with an overall recognition accuracy of 92.65%, a recall rate of 93.63%, and an F1 score of 90.47%. The model demonstrates relatively high performance across various precision metrics for knowledge extraction and exhibits good robustness. In knowledge-based question-answering, the model successfully extracts relevant information, achieving a match rate of 58.81%. The established knowledge graph promotes the inheritance and development of Chinese classical culture in international Chinese language education.