Research On Automatic Question Answering Technology In Restricted Domains

With the rapid development of Internet, the resources of online information has become increasingly diverse. The number of web pages has been more than 10 billion worldwide,and it increases in the number of millions per day. Network has become one of important way to access to information. In order to quickly find the page from the mass of information,people should take advantage of search engines. In the absence of combination with the semantic understanding,it is difficult to analyse the real intentions clearly in question answering. Especially for Chinese information search,the results are often unsatisfactory.The subject proceed researching and testing within the field of real-name obtaining information system. The automatic question answering system is based on the existing search technology-keyword matching which widely use, added semantic matching words, analysis of questions structure and syntax, and overcomed the shortcomings which exist in the traditional search engines, lacking the understanding of semantic information and reasoning ability between knowledge relationship,which is helpful to enhance the accuracy of answers.The subject combines theory research and practical application to propose the method of integrated keyword matching, semantic analysis,parsing. It adopt building FAQ database, building domain ontology library, and semantic matching and logical reasoning on the basis of building domain text library in order to realize intelligent question answering. In the process of questions anlysis, with the help of natural language processing technology can get the intent of questions fast and accurately, and using a variety of strategies to answer extraction.The research use real-name obtaining information system as the study field, and the method has been approved.The practice results of QA system development prove that the search technology integrated semantic analysis and parsing is not only feasible but also reliable.It is fast and able to return more accurate answers to users' needs. Meantime, It has summarized out the method of improving search in the practice of research, development and application.Therefore,it has laid a good foundation for further research.