A Question-answering System Based On The Domain Ontology And Sentence Template

With the development of computer science and Internet, people increasingly query the information on the web with search engines. But the search engines exists many drawbacks. First, search engines return a large amount of documents that include indexing terms in a user’s query, and users have to further select the information. Second, the combination of several key words is difficult to clearly express the user retrieval intention. As the same time, it is hard to improve the retrieval effect, because the matching algorithm based on key words index does not involve semantic. Therefore, it’s important to explore more humanity and efficient search engine. And the question-answering system is the collection of knowledge representation, information retrieval and natural language processing. It can better satisfy the demand of the user search and has become a new research hotspot.On the other hand, the distance education is accepted by more people with the development of internet technology, and its outstanding characteristics are the separation of space and time, teachers and students lack of effective communication. The intelligent question-answering system combines with the natural language processing to understand the user’s question with natural language from, and gives a concise, clear answer to users. Thus, the intelligent question-answering system can solve the problem of student questions in the distance teaching.At present, Chinese question-answering, based on natural language understanding, have low accuracy rate which main difficulties lie in the construction of the knowledge base and computer understanding the semantic of the user questions. The question-answering system in our paper bases on ontology knowledge base and sentence templates. First we construct the domain ontology knowledge base using the ontology technology, use the constructed concepts and relations in the Tbox to define the abstract concepts during building the ontology. The definition of abstract concept overcomes the difficulties in expressing the complex concepts which excellent satisfies the user’s question based on concept definition and characteristics. Second, the paper selects the sentence template matching algorithm based on semanitic block. The same Semanitic Template was added into sentence template basa. The theme semanitic block, distinction semantic block, question semantic block, assist semanitic block consist the question template. In the papper, it formulates rule of distinguishing the semanitic block, and gives the method of calculating template similarity. The improved sentence template method reduces the demand of the template number, and has great advance in the ranges of dealing with question and accuracy of understanding question. Then, we design the question-answering system model based on domain ontology and the improved sentence template algorithm. The model has been improved in the question understanding, automatic word segmentation and the ontology knowledge base which greatly increases the accuracy of question-answering system. Finally, paper implements a Chinese intelligent question-answering system faces to computer foundation course.The pivotal of this paper consisted of four issues as follows:First, introduce the basic knowledge of question-answering system. The paper analyses the study situation of question-answering system at home and abroad, the meaning, classification methods, and main contents of the question-answering system. It also discusses the difficulties in the Chinese automatic question-answering system. According to the key of this paper, we mainly introduce the contents of the ontology and domain knowledge, and the related Chinese language processing technology including the existing word segmentation algorithm and the methods of understanding question.Second, construct the ontology knowledge base. Ontology knowledge base is benefit to organize, manage, maintain, reason and query the knowledge. In this paper, the biggest characteristic of the ontology knowledge base is the definition of the abstract concepts based on special relation. We use the constructed concepts and relations in the Tbox to define the abstract concepts during building the ontology. The definition of abstract concept overcomes the difficulties in expressing the complex concepts which excellent satisfies the user’s question. Compare to those different methods, we select the knowledge engineering method to build the domain ontology knowledge base faced to computer foundation course.Third, understand the semantic of the user questions. The paper selects the sentence template matching algorithm based on semanitic block. The same Semanitic Template was added into sentence template basa. The TZB, ZTB, YWB, FWB consist the question template, and only the TZB and ZTB are relation with the semanitic information of the user’s question. With the characteristic of semanitics block, the papper formulates rule of distinguishing the semanitic block, and gives the method of calculating template similarity. And through a set of test that confirms the improved algorithm increases the accuracy of understanding question in a certain degree.Fourth, the design and implement of question-answering system based on ontology knowledge and sentence templates. The model has been improved in the question understanding, automatic word segmentation and the ontology knowledge base. The structure of the system is made up of foreground natural-language interface and background ontology knowledge base. Through the improved sentence template algorithm, the system acquires accurate semantic of the question, adopts the corresponding query from to search and reason ontology knowledge base according to different semantic information, and reorganizes the searched results. Then the organized answers will be sent to the user. Last, the paper implements a question-answering system facing to computer foundation course via the constructed ontology knowledge base and the improved sentence template matching algorithm. This system not only can query individuals, but also can answer the question about definition and characteristic of the concept, which can be applied in the teaching domain to solve students’question.