Design And Implementation Of Books And Movie Question Answering System Based On Knowledge Graph

People's demand for film and television and book information is growing,but at present,the way for users to actively seek film and television and book information is generally relying on search engines.Traditional search engine based on keyword search,the results are not satisfactory,question answering system has become the solution.There are a large number of unstructured and semi-structured data on the network,which does not form structured available knowledge,while knowledge map can transform unstructured and semi-structured data into structured knowledge,and question answering system can use natural processing technology to accurately answer user questions,because the combination of knowledge map and question answering system is widely concerned by academia and industry.Based on natural language processing,knowledge map and other technologies,this paper designs and implements a knowledge map Q & a system for film and Book fields.The main work of this paper is as follows:Firstly,the related concepts and construction technologies of knowledge map are described.Then,the data about film,TV and book information is acquired by web crawler.The technology of knowledge map is studied to transform unstructured and semi-structured data into structured data.According to the mapping relationship between data analysis and design of knowledge map,it will be imported into the map database neo4 j,so as to build a knowledge map for the field of film and television and book information,to provide data support for the Q & a system.Then,three problems are solved in the construction of automatic question answering model.For the question classification,thirteen types of questions are constructed,and a naive Bayesian parallel algorithm based on spark is designed to classify the query questions input by users,so as to improve the classification efficiency.For the extraction of question key words,in view of the serious gradient explosion and disappearance of RNN in the process of forward and back propagation,which will lead to the disappearance of long text information in the training process and the poor fitting effect of the model,a method based on the improved long-term memory network(LSTM)and conditional random field(CRF)is designed to extract the text words Semantic information can avoid the problem of gradient di sappearance and gradient explosion,so as to improve the effect of keyword recognition.In view of the diversity of question input,a fuzzy query method based on the combination of two-way segmentation matching algorithm and cosine similarity algorithm is designed to improve the effect of question answering.Finally,this paper realizes the prototype system of book and movie Knowledge Q&A based on knowledge map,gives the system structure chart,introduces the design and implementation of three modules: knowledge map construction,question processing and system interaction,and tests and displays the Q & A effect.The system can recognize the key words of questions and classify them,convert them into cypher query statements,and get accurate answers from the database of knowledge map.