Research Of The Key Techniques Of Agricultural Automatic Question Answering System Based On Ontology

With the development of network technology, access for information has become animportant part of people's daily life. Therefor nature language processing technology hasbeen promoted and also has been put forward for a higher request. How to acquire servicesrapidly and accurately in the world of network is a difficult research. It is in this responseautomatic question answering system has developmented. Automatic question answeringsystem is an important form of information search, and it is proposed to help retrievingwith higher efficiency than traditional form. Currently, the question answering system isstill at an exploratory stage. It has been constrained by low level of natural languageprocessing and AI. The existing question answering system focus on the search forkeywords of the users' questions, semantic understanding, the most important of thesystem has not been involved.To solve these problems of above, this paper regards field ontology as the foundationof semantic understanding. Ontology is a kind of modal which is used to describe theconcepts and the relations of them. It includes abundant knowledge and semantic relationsof this field. Through applying these resources to question answering system by a certainway, it is impossible to solve the problem of semantic understand insufficiency in a way.The article, by constructing a small ontology to assist in the understanding of thesentence, search and the calculation of the semantic similarity between sentences.Determine the scop of the ontology, get the concepts, define the attributes of concepts andthe mutual relations between them, owl used to describe the ontology and then ontologystored in such a manner as determined. The question answering system for module designand build question-answer bank, and then analyse the questions based on ontology. Get thesentence pattern as well as the center concept and the concept set. Design search programs,search candidate question set. Candidate question set will be set with the user's questionfor the calculation of the semantic similarity and then the result which higher than thethreshold value be returned as the answer.Now the phrase similarity algorithm based on the statistical method or the semanticdictionary is tend to be used, the former leave out of account the semantic information ofwords and the latter is subject to the limitations of the dictionary itself. So the article usethe semantic similarity calculation method based on ontology, and discussed the semantic similarity calculation model between the concepts and between the sentences. Thesimilarity algorithm between concepts is mainly by the contrast of various attributes ofdifferent concept meanwhile more similarity calculate across semantic neighborhoods. Toensure accuracy in the calculation of a large number of concepts, the paper introducedtaboo algorithm to find the optimal solution. At last design the Vector Space Model ofquestion, and calculate similarity between sentences draw on the model. Finally introducethe system prototype and analyse the result of the experiment to prove the validity andeffectiveness of the algorithm.