Question And Answer System Based On Ontology Design

With the rapid development of network technology, remote education has been paid more and more attention. In the remote education mode, teachers and students are separated from each other and can not communicate directly. The students have to study by themselves. As one important part of remote education, question answer system on web makes it easy for students to answer their questions and get rid of the barrier in their study. It plays an important role in the communication between teachers and students, and in helping students to get information. It can not be replaced. So the research of question answer system on web is very important to improve both the remote education and the education in China.There are many deficiencies in the existing question answer systems on web. Some question answer systems perform the asking and answering process by Email. Though in other question answer systems, some new technologies such as key words matching, segmenting and searching are used, they are still short of semantic understanding. So the effect is not very well.We propose an ontology-based question answer system on web in this paper. A domain ontology is built as the foundation of semantic understanding. The effect of semantic understanding, share and reuse of knowledge is good by using this technology. We will show the new method to build the domain ontology, which is useful in question model matching and similarity computing.Building a domain ontology is the key part of question answer system on web. We describe this process as follows: making the domain and area clear, picking up the concepts, making a definition of the relationship of the concepts and giving the form of the domain ontology,etc.Since the questions are various, the question will be segmented to get a set of key words. The mixed method based on character matching and frequency counting will be used.Referring to the information in domain ontology and question model store, we make semantic analysis to get the models of questions. Then, the transition network is used to search for the answer.For the questions which are not normative, we use both the Vector Space Model method and the semantic-based similarity compute method to match the answer.In conclusion, the ontology-based question answer system on web uses the domain ontology effectively. It strengthens the semantic understanding in question answer system on web and improves the accuracy of the answers.