Named Entity Recognition Based On BiLSTM-CRF

As the basic content of natural language processing,the task of named entity recognition is to identify the entity with specific meaning from the text to be processed,and the recognition result has strong relevance to the follow-up work of natural language processing.The main task of Chinese named entity recognition is to identify common entities such as person name,place name and time in Chinese text,and the accuracy of name recognition is particularly important in professional fields,such as information extraction in network attack and defense field,intelligent question answering system,emotion analysis,etc.Named entity recognition is a basic and key step in natural language processing in these professional fields.The difficulty of using traditional methods for named entity recognition is that it relies heavily on rule matching and manual feature extraction,and requires professional domain knowledge to achieve better recognition effect,and needs to maintain the correctness of rules according to the change of situation for a long time,so it consumes a lot of human resources and has high time cost.With the development and progress of deep learning and the continuous growth of unstructured text data,named entity recognition based on deep learning gradually highlights its importance.In this paper,a bi directional long short term memory neural network(BI LSTM)combined with conditional random field(CRF)is proposed to improve the accuracy of named entity recognition.Firstly,the text character level vector with label is input into the model,and then the entity is graded according to the semantic relationship in the context.Then,the prediction label of the entity is determined in the CRF layer according to the restriction relationship between the tags.Then,the model training is used to reduce the error between the correct label and the prediction label,and improve the accuracy of the model.The experimental results show that the accuracy,recall and F1 score of the model are90.1%,95.4% and 93.5%,respectively.Compared with the traditional named entity recognition method,the effect is significantly improved.