The communication between the teachers and students is very important in network education. Question-answering system can resolve the problem. This paper proposes a question answering system in network education based on FAQ and then introduces the kernel module which is the calculation of the question similarity in detail. The paper presents the similarity algorithm of domain keywords and common words respectively and integrates them into the question similarity. Experimental results show that the proposed method can achieve good performance and the system is applied.

In Chinese question answering system, because there is more semantic relation in questions than that in query words, the precision can be improved by expanding query while using natural language questions to retrieve documents. This paper proposes a new approach to query expansion based on semantics and statistics. Firstly automatic relevance feedback method is used to generate a candidate expansion word set. Then the expanded query words are selected from the set based on the semantic similarity and semantic relevancy between the candidate words and the original words. Experiments show the new approach is effective for Web retrieval and out-performs the conventional expansion approaches.A typical question answering system is made up of four modules which are question analysis, document retrieval, candidate answer extraction and answer selection. When using question presented with natural language to retrieve document, query words extraction and query expansion are the key factors that affect its retrieval performance. After stop-words are removed from Chinese question segmentation, nouns, adverbs and adjectives are extracted to form the query. The query is submitted to search engine to retrieve relevant documents. The object of information retrieval is to provide the user with a set of documents that are relevant to the query, at the same time to eliminate the irrelevant ones from the retrieved set. Currently, a fundamental problem in document retrieval is term mismatch between queries and documents [1] .With the explosive growth of information in Internet, the word mismatch problem in a more general context becomes more significant as observed by Furnas [2] . In his experiments, the probability of two users using the same term to describe an object is less than 20%. Addressing the word mismatch problem has become an increasingly important research topic in IR. For query already generated, it is necessary to expand query. A query is expanded by adding other terms closely related to the original query terms. In the past years, there are three kinds of query expansion methods in information retrieval. The first is automatic relevant feedback [3][4][5] , which statistically analyzes the initial retrieval docu-

No abstract

No abstract