Chinese Named Entity Recognition in Industry Domain Based on Lexical Enhancement
Jingxin Zhao, Zhenxing Ni, Cuilan Liu, Jiadong Ren, Caiyun Liu, Yuqing Zhang · 2024
Named entity recognition is the premise of realizing information extraction and information retrieval in industrial domain, and it is also the basic means of realizing industrial data security management. As an important basic industry in the industrial field, the steel field has huge data, complex text semantics and numerous proprietary terms, so it is difficult to identify named entities in steel domain because of the complex semantics of the text and the variety of proper nouns. In order to improve the performance of named entity recognition in steel domain, a named entity recognition dataset of steel and a MacBERT-FLAT-CRF model based on lexical enhancement are constructed. The results show that, compared with other NER models, this model has a good effect on the self-built named entity recognition dataset, which effectively improves the recognition effect of named entities in the steel field and improves the problems of insufficient use of vocabulary information and lack of data in the steel field.