Research On Named Entity Recognition Algorithm Based On Attention Mechanism

With the rapid development of the Internet,the information on the Internet is becoming more and more abundant.It means that it is more and more difficult to find available information quickly and efficiently in massive data.There is an urgent need to extract useful structured data from massive unstructured text data.For this reason,information extraction technology has emerged.As one of the important tasks of information extraction technology,named entity recognition has always been one of the focuses of researchers at home and abroad.Based on this background,this paper studies the named entity recognition.Specifically,it is divided into two aspects.One is the study of named entity recognition in English named entity recognition,and the other is the research in Chinese named entity recognition of legal verdict.The main contributions of our work can be summarized as follows:1.In the field of English named entity recognition,we strengthen the semantic relationship between named entities and tags by aligning named entities with tag semantics.We propose a word-tag semantic alignment model.Notably,an attention mechanism is proposed to capture the semantic relation between word and tag,and make the predictions better because of the semantic alignment of word and tag.We design a supervised attention mechanism to supervise the semantic alignment of word and tag.2.In the field of English named entity recognition,we design a joint model between named entity recognition and language model.the joint model is jointly trained to enhance the semantic representation of text sequence and improve the results of named entity recognition.3.The Chinese legal verdict is used as the experimental data,and we study named entity recognition in the criminal facts.Extracting the criminal facts part of the judgment by rules,and learning the word embedding by Word2vec.We design and implement a recurrent neural network model based on attention mechanism to learn grammar information in named entity context,and learn more context information by introducing a bias objective function.