Joint Entity Recognition And Relationship Extraction Model Based On Deep Learning

Entity recognition and relationship extraction are two very classic problems in the field of natural language processing.The ability to quickly and accurately identify entity pairs and their semantic relationships is essential for information extraction,and is of vital significance in the process of creating knowledge graphs and information search.With the rapid development of the Internet,the amount of data is increasing day by day,and the requirements for knowledge services are becoming higher and higher.Therefore,the entity relationship extraction technology has also become one of the research hotspots in recent years,and it has played a great role in many fields such as information retrieval,question answering systems,knowledge base creation,and knowledge graphs.With the development of deep learning,the advantages of neural network-based entity recognition and relationship extraction technology have been fully revealed,and gradually become one of the current mainstream methods.This article studies the joint entity recognition and relationship extraction model based on deep learning.The main work of the article includes the following aspects:1.Summarized the research background and research significance of entity recognition and relationship extraction as two hot tasks in the field of natural language processing,introduced the principles and quality evaluation indicators of entity recognition and relationship extraction tasks and reviewed the development of entity recognition and relationship extraction History and current research status at home and abroad.The basic theories related to deep learning and neural networks,such as word vector technology based on neural networks,convolutional neural network structures,recurrent neural networks,long and short-term memory networks,and optimization algorithms,are introduced in general.2.A joint model for deep learning entity relationship extraction based on bidirectional long-short memory network is proposed.The model uses a two-way long-term and short-term memory network to encode the language context of the entity.The purpose of joint entity recognition and relationship extraction is achieved by sharing parameters.The semantic information is deeply used and error propagation is reduced.The experiments on the Co NLL04 dataset and the COAE2016 dataset show that the model proposed in this paper is significantly better than multiple benchmark models.3.An end-to-end model based on remote supervision is proposed to perform entity relationship extraction tasks.The model also uses a two-way long and short-term memory network to encode the language context in which the entity is located.A new labeling strategy is used to completely convert the entity recognition and relationship extraction tasks into sequence labeling problems,and then add a self-attention layer to fully represent the text.Through the loss function with offset term,the model's ability to identify related entity pairs is enhanced.Finally,the experimental results on the NYT dataset prove the effectiveness of the model proposed in this paper.