Chinese named entity identification method and Chinese named entity identification device based on RoBERTa-BiGRU-LAN model

Abstract

The invention belongs to the technical field of named entity recognition, and particularly relates to a Chinese named entity recognition method and device based on a RoBERTa-BiGRU-LAN model, and the method comprises the steps: converting a to-be-processed Chinese corpus into a word vector sequence; inputting the obtained word vector sequence into a first layer of BiGRU-LAN of a RoBERTa-BiGRU-LAN model, and obtaining a coding sequence fused with local information; inputting the obtained coding sequence into a second layer of BiGRU-LAN of the RoBERTa-BiGRU-LAN model, and obtaining attention distribution fused with global information; and obtaining a named entity identification result according to the obtained attention distribution. According to the improved word embedding method disclosed by the invention, Chinese representation is better carried out, and meanwhile, BiLSTM-CRF is improved into BiGRU-LAN, so that the parameters of the model are reduced, the complexity of the model is reduced, and the training time is saved.

Description

technical field [0001] The invention belongs to the technical field of named entity recognition, and in particular relates to a Chinese named entity recognition method and device based on the RoBERTa-BiGRU-LAN model. Background technique [0002] Entity is an important part of carrying semantic information in text, and is the core unit of knowledge graph. Named Entity Recognition (NER) aims to extract these valuable entities (person names, place names, institution names, proper nouns, events, etc.) information from text to meet the needs of various industries. Named entity recognition is one of the key steps in the field of natural language processing, an important basis for building knowledge graphs, and one of the core technologies in the fields of intelligent search and intelligent question answering. It is of great significance to realize tasks and realize artificial intelligence supported by knowledge. [0003] The current Chinese named entity recognition method widel...

AI Technical Summary

Problems solved by technology

[0004] (1) The word2vec model is a static word vector model, which cannot solve the problem of polysemy and polysemy. The difference of different meanings will interfere with the final named entity recognition effect

[0005] (2) When using the traditional Bert model for word embedding, due to the use of static masking, small training data sets, and insufficient training time, representation learning is insufficient; on the other hand, the model optimization rate and model performance using Bert are weak

[0006] (3) Compared with the traditional RNN, the BiLSTM model has too many parameters and the model complexity is higher

[0007] (4) CRF does not add other additional information to the sequence, and the computational complexity is high

Images