Research on Ancient Weapons Named Entity Recognition From Chinese Historical Novels

Abstract

As a basic task of natural language processing, named entity recognition has been applied in the field of Chinese literature. A method can be realized the ancient weapons named entity recognition and association analysis in Chinese historical novels,so this paper puts forward the recognition model and use knowledge mapping analysis. Taking ancient weapons in Romance of The Three Kingdoms as a characteristic entity example constructed a corpus,the BERT-BiGRU-CRF model is built to solve the named entity recognition of ancient weapons based on digital humanities.And the knowledge map is used to analyse the relationship of entities.[Result]The training effect of the combined model BERT-BiGRU-CRF is obviously better than the other four models, and finally the harmonic average F1-score of the ancient weapons’ entities is improved to 88.44%.This method can solve the difficulty of recognizing the ancient weapons of Chinese historical novels, providing reference for the further entity correlation of Chinese historical literature.