Named Entity Recognition in Judicial Field Based on BERT-BiLSTM-CRF Model

Abstract

The recognition of named entity in judicial documents is the key to realize automatic trial, and how to effectively distinguish the entities from text is the focus of this paper. However, in special fields, such as the judicial field, many experiments show that the artificial features selection based on domain knowledge have a great influence on the results of the neural network models. Therefore, how to obtain a better named entity recognition performance in judicial field without relying on artificial features is a problem to be solved. In this paper, we propose a neural network model based on BERT-BiLSTM-CRF. Firstly, we use the BERT pre-trained language model to generate the word vectors according to the context of the words, enhance the semantic representation of words, then the word vector sequence is input into BiLSTM-CRF for training. Experiments show that our method is effective, at the same time, it solves the problem that the traditional word vector representation maps the word into a single vector and cannot characterize the ambiguity of words.