Frequently Asked Q & A Library Of Chinese Question Answering System Based On The Restricted Domain

Question answering system provides the human-machine interface by means of natural language. Comparing to the traditional search engine which is based on keyword, question answering system has prominent advantage. In the restricted domain, question answering system on the basis of frequently-asked question is more accurate, simply and efficiency at localization to the sentence-answer, in each domain of daily life, it is the key problems of the research and will be a brilliant application in the future. it is a hot problem of present research. This paper does a series of research which revolve method of a knowledge base constructed in the domain, method of sentence-answer retrieval, the collection and organizing of FAQ in the restricted domain, question classification, retrieval of related question sentences from the candidate question base and extraction of question answers in the implement process of FAQ question answering system. The main innovative achievements are as follows:(1) This paper puts forward a method to construct a knowledge base in the domain in restricted domain., which utilizes the characteristic of domain knowledge, the fusion of domain knowledge base (Domain HowNet) and common knowledge base(HowNet) is realized with the help of the idea of ontology and common knowledge base(HowNet),and more effective resources for research of natural language processing is offered.(2) This paper presents a method of making use of the characteristic of field knowledge to collecting, organizing and sorting out FAQ. This method implements the collecting, organizing of FAQ in the basis of analyzing the characteristic of travel field according to the characteristic of region and question type.(3) This paper brings forward a method of classifying and identifying domain question based on rules. This method utilizes the language rule of the question sentence and the characteristic of domain knowledge to extract the classification rule of the question sentence, and classify and discern the question sentence on the basis of the rule. This method has a very good effect on dwindling the candidate question base and improves the accuracy of the location of the answer.(4) This paper presents a semantic dependence question sentence similarity algorithm utilizing the characteristic of domain knowledge. This method makes use of the characteristic of domain knowledge to extract the classification of the question, and filtrate the similar question according to the classification. Based on "HowNet" and "Domain HowNet", it makes use of the syntax analysis to extract the efficient dependence pair and through the relation between dependence pair and concept semantic to implements similarity calculate among questions.(5)In the YunNan travel domain, the YunNan travel question answering prototype system are designed and implemented. "Domain HowNet" and domain FAQ database are constructed, implements question answer retrieval by method of question sentence similarity calculation in the paper. The experiment result of YunNan tourism question answering model shows that the method proposed in the paper is feasible and effective.