Phone Domain Knowledge Base Question Answering System Based On Named Entity Recognition

With the development of the information industry,mobile phones have become popular in our life.There are so many kinds of smart phones with so many functions that users will encounter many problems before and after sales.Enterprises usually solve such problems by providing intelligent customer service and manual customer service,which brings huge cost pressure to enterprises.The traditional customer service system can only provide preset functions for the user to choose from.At this time,a question answering system that can truly understand the user is needed to alleviate such problems.Knowledge base question and answer is a question and answer system which is given a natural language question,through semantic analysis in various ways,and makes use of knowledge base to make inquiry and reasoning to get the answer.With the development of deep learning,more and more people pay attention to it.But the industry has not yet a full-fledged answering system for mobile phones.In this paper,a knowledge base question-answering system is designed and implemented.The main work of this paper is as follows:First,the knowledge base construction of the mobile phone field is completed.This paper uses the method based on grammar analysis and syntactic analysis and combines some external data to realize the extraction of related entities in the mobile phone field,mainly including mobile phone attribute entities,mobile phone function entities,mobile phone fault entities and mobile phone name entities.It is stored in the Hbase database.Secondly,the mobile phone domain named entity recognition service system is implemented:Bi-LSTM + CRF model is used to train the mobile phone domain named entity recognition model.Flask is used as the Web service framework,which provides named entity recognition service to other modules.Finally,after testing,the named entity recognition service reached over 97%F-measure.Thirdly,the question answering system in the mobile phone domain is constructed:the system calls the mobile phone named entity recognition service,and analyzes the most critical semantic information of the text.Combining with the text classification model of TextCNN,the question analysis is realized,and the final answer is obtained by querying the knowledge base.Using Redis cache technology and docker container technology to deploy multiple modules using Nginx for load balancing,the pressure of requests can be distributed to multiple servers to reduce the pressure of single point service.After developing and testing,the QA system can achieve 87.2%accuracy.The domain knowledge base implemented in this paper has been put into practical use,which is convenient for enterprises to model and analyze user FAQ data.The research and development of this system has played a certain role in promoting the development of the research field of knowledge base question answering.