The Study On Question Recommendation Technology In Community Question Answer System

With the development of web2.0 technology, appearing more and more applications to help the users area exchange ideas and share knowledge. In the web2.0 platform, community question answering system become the preeminent, attracting more and more users involved. It makes up for the traditional search engine can provide users with a simple search results, to provide users with an interactive experience, also makes up for the traditional question answering system is simply answer the question of the machinery, and there is no better for the user to provide better service. For each user in the community question answering system, they are all free, free to interesting questions, at the same time can also to answer his concern, can also be interested in their problems are discussed, the powerful interactive promoted the rapid development of the community question answering system. With the rapid development of the community question answering system, community question answering system become the important platform of knowledge acquisition and sharing.First, Since the community questionnaire system user has author and issues respondents question the dual role, this paper proposed a fusion user questions recommended method of dual role and user questions relationship. This method is mainly using PLSA model based on user history and answer the questions of the problems and the problem to analyze the dual role of users in the system. At the same time the user questions relation model based on language model to in-depth mining the relationship between the user and the question.Then, due to the nature of the recommended questions that provide the user with similar problems and raised questions, first put forward a question recommendation method based on user interests and needs, mainly using PLSA model based on user history records of answering the question to find the user’s interest. At the same time, based on the translation model based on the user’s query. The two methods respectively in the Yahoo! Extraction of real Answers website annotation data set on the test, and through the multiple perspectives of comparative experiments show that the proposed model has achieved good performance.