Behavior Analysis System Of SNS Users Based On Data Mining

With the information technology improve our lives gradually, Many social networking services are published, such as RenRen netword, KaiXin network, Tencent Friends network etc. They provide people with the integration of knowledge, consultation with the problems, exchanging friendship and other new features. Through the analysis of user behavior, they provide different users specific services,which will be able to greatly attract users. The purpose of this article is to build a personalized smart recommendation engine by analyzing the characteristics of community site users. This system could get the potential of user interest, and fully enhance the users experience.By reference to the papers about data mining and behavioral analysis, this article first design the general framework and its key business processes based on other behavior analysis systems. And then we design characteristics collection, features of pretreatment, data mining algorithms, and feature data indexing detailedly. At last, we described a time scheduling mechanism for this system.To efficient behavior analyse, this article refer to many research results, we improve regular expression matching algorithm to pre-process data efficiently. And we convert the problem of user behavior analysis to a ranking problem, which could use Ranking algorithm to do data mining. At last this system will dig out the data characteristics, and map into a high dimensional space by using LSH algorithm,which has a high-performance.In experiment, the text segmentation engine could analyse the users'input text quickly, and with high accuracy. And the regular expression multi-mode matching algorithm has been optimized to reduce memory consumption. By testing improved LSH algorithm, it can efficiently establish the index of the characteristics of user interest. Therefore,this system could provides a feasible scheme. for community website user behavior analysis.