Research And Implementation On Ontology-based Domain Question And Answering System

With the rapid development of network technology, Network has become the main places where people can obtain information and communicate with each other in nowadays and future. And the whole network has become an enormous, dynamic information service network with multiple information resources and services around the world, and provided valuable information resources for the users.In order to satisfy the requirement of people who want to obtain valuable information on network, lots of search engines came into people's lives. Being an important kind of search engine, compared with traditional search engines, Question and Answering System is able to obviously improve retrieval quality, but in terms of this system in home and abroad, the feedback of the system is not very satisfying. The main reasons can be listed as follows:First of all, Question and Answering System needs to proceed grammatical parsing for a question sentence, but for now, accurate analysis to a sentence is too difficult;Second, While comparing the differences and similarities, the current Question and Answering Systems only take into account keywords matches between the sentences, while ignore the semantic relationships between them.In order to solve these two problems, Ontology is introduced into this Question and Answering System, as a result that it can do semantic analysis and similarity calculation to the sentences by using the knowledge described by ontology, which can make the result more extensive and accurate.To sum up, the main work of this article can be listed as follows:①Avoiding complicated Chinese lexical and grammatical parsing , based on the ontology, a Chinese sentence can be divided into ontology dimension and common dimension, consequently, it is not only simplified, but also its original meaning never be changed.②Classify the questions of Frequently Asked Questions based on the hierarchy information of ontology. Thus some useless sentence can be filtered before the similarity calculation; as a result the efficiency of system has been improved.③With employing the information described by ontology, the keyword of users'questions are proceeded expansion, consequently, precision radio of system can be improved. ④A novel method about similarity computation is proposed based on ontology. In similarity calculating we both considered the ontology dimension and common dimension of a sentence, the similarity between sentences should be weighted sum of them.⑤An Ontology-based Domain Question and Answering System (ODQAS) have been implemented. And the ultimately experiment result shows that the similarity computation between sentences proposed in this article is validate, additionally, compared with other systems, this system's performance have a good robustness.