Research On Community Question And Answer User Interest Modeling Methods

With the application and development of Web 2.0 technology,as a typical application in the new knowledge acquisition model,Community question answering(CQA)enhances the interaction between questioner and respondent,and makes the connection between users more closely.At the same time,users can search and evaluate the existing knowledge and information in the Q&A platform,compared to the search engine,this application is undoubtedly a new information acquisition model,and it's more convenient and effective for users.However,with the continuous development of the CQA,it attracts a large number of users to participate,which makes information resources become more and more large in the community,and because of the unique community interaction model,making CQA face many problems and challenges,then how to accurately understand the user behavior and needs,to provide users with personalized recommendation service has become an urgent problem of CQA must to be solved.User interest modeling as a key technology in personalized recommendation service,the use of user interest modeling is directly related to the effect of personalized recommendation service.This paper analyzes the related problems of user interest modeling in the CQA,and mainly completes the following work:(1)Take“Baidu knows”as an example,we have a specific analysis of the structure in the community question answering.Before modeling the interest of CQA user,we need to understand the user's environment.Through this analysis,we can understand the interaction of user in the community,the community asked the needs of users with a certain awareness,and it provides a certain basis for our better access to user interest information.(2)For the special attributes of CQA user's tags.This paper considers that the user's tag in the community question and answer is generally associated with the type of problem,can to a certain extent,reflect the user in a field of knowledge of the interests or authority.So in the user interest feature extraction process on the integration of the role of the label,it can effectively solve the sparseness of the user's interest feature set and enhance the user's interest characteristics.Finally,we build a community tag user interest model with a fusion tag.And the spectral clustering algorithm is used to study the model in the modeling process.The experimental results show that it can effectively improve the accuracy of the user interest modeling after the fusion tag characteristics,and the spectral clustering algorithm is superior to the traditional k-means clustering algorithm in model learning.(3)For the interest Offset of CQA user,we analyzed the characteristics of user interest changes.Based on the theory of Ebbinghaus' forgetting law,using the interest of the forgotten function of the user interest to do the time optimization update calculation.And finally constructs the community question and answer user interest model based on the forgotten rule.The experimental results show that the method is better than the fusion tags in the modeling method of the CQA user.