The Research And Implementation Of The Technology Of Chinese Question Answering System Based On Restricted Domain Ontology

With the development of Internet technology, information explosion has a great change in people’s lives. Nowadays, people are becoming accustomed to obtain a variety of information on the Internet, which is mainly due to the development of the search engine technology. However, traditional search engines still have some drawbacks. On one hand, Information is only able to be retrieved by keyword, which does not adequately express the intent of users. On the other hand, many candidate answers are returned by search engine and user have to spend some time in finding their targets, in which the user experience is usually not good enough. To solve the problems above, question answering system is raised. Questions expressed in natural language are submitted to automatic question answering system and the answers of the questions are directly returned.First of all, we analyze the technology and theory of current existing question answering systems and expound the advantage and disadvantage of the technology used in the corresponding modules. Secondly, referring to some ontology construction works abroad and following the methodology and experiences of this ontology engineering, we build a small-scale ontology based on retail domain which is used in the retrieval of information in restricted domain knowledge. Thirdly, the answer extraction method of question answering system based on retail domain is raised. Questions expressed in natural language are submitted to the system by users, and then are segmented, eliminated stop words, annotated semantically. Through the step of shallow semantic parsing, the known and unknown information in the question are obtained, on the base of which question vector is generated. Last but not least, we use SPARQL to find the answers to questions from the ontology library. And then the answers are directly returned, which increase the recall rate of the system and improve the user experience. Based on the theory above, we design and implement the question answering system orienting retail domain. Through the application system, the related theory proposed in this paper is verified and the feasibility is also proved.