Intelligent Question Answering System Based on Knowledge Graph of Beijing Opera

In order to improve the digital service effect of traditional Chinese culture, this paper focuses on the field of Beijing Opera, completes the research and construction of the knowledge graph of Beijing Opera, and further builds an intelligent question-answering system. For Chinese questions input by users, the system uses the BERT + Bi-LSTM + CRF model to recognize named entities, then uses the BERT-Sentence Pair Classification model to calculate text similarity to complete relational attribute mapping, and combines character string matching to improve efficiency, finally gets the answer. On the Beijing Opera data set of this paper, the F1 values of the two models are 91.01% and 91.17%. Finally, this paper proposes an ontology construction method of the domain knowledge graph, the built question-answering system has also achieved good application effect on the retrieval and utilization of Beijing Opera knowledge, and realizes an effective form of digital cultural services.