Research And Application Of Key Technologies Of Domain-oriented Question Answering System

With the emergence and rapid development of the Internet,there are more and more data resources on the Internet,and it is becoming more and more difficult for users to obtain true and reliable information from massive Internet data.At present,the main method for people to obtain resources is through the search engine,but the information returned by the search engine to the user is often redundant,requiring the user to conduct a second screening to obtain their own information,which greatly wastes the user's information time.Therefore,more and more research institutions and technology companies have begun to deeply analyze and develop intelligent question and answer robots,such as Apple 's Siri system,Microsoft 's Microsoft Xiaobing,and Ali 's Ali Xiaomi.The core module of the intelligent question and answer robot is question answering module,therefore,automatic question answering system has become a research hotspot in the field of artificial intelligence.This article first analyzes and introduces the relevant knowledge mainly technologies in the field of question answering system,and finally implements a question answering system in the field of party building.The system is based on the domain knowledge base and contains core modules such as domain knowledge crawling,retrieval question answering and knowledge graph.The domain knowledge crawling module crawls the domain website to generate domain term collections by manually setting the seed dictionary,and then crawls the knowledge of the encyclopedia website and the community question and answer website through the term collection.The retrieval question and answer module uses the method of information retrieval In the implemented question answering system,the module first constructs an inverted index to retrieve candidate answers,and then uses the semantic matching algorithm to reorder the candidate answers,and returns the sorted answers to the subsequent answer generation module.This paper explores and analyzes the Chinese question matching algorithm,and proposes a semantic matching algorithm based on the siamese neural network model.The experimental results show that the performance of the algorithm is better than other algorithms.The question and answer module of the knowledge graph stores knowledge in the form of a semantic web.The natural questions entered by the user are converted into SPARQL query statements to retrieve knowledge,and the results are returned to the subsequent answer generation module.This article starts with machine translation and implements the generation algorithm from natural language to SPARQL query statements.The answer generation module screens the retrieved answers and the knowledge graph answers,and finally returns the results to the user.The experimental results prove that the accuracy rate of the party building question answering system designed in this paper reaches 91.46 %.