A Method Of User Interest Mining In Micro-blog Based On Behavior

As a newborn network application of Web2.0 era, microblog has led to an unprecedented so-called micro-revolution because of its amazing influence as well as the rapid rise in the number of users shortly after its birth. Different from other social networks, microblog is not only a medium of communication and information dissemination,but its amazing message diffusion also has a broad application prospects in advertising marketing and personalized service. While enhancing the user experience and increasing the satisfaction of microblog users, accurate advertising and friends recommend also bring huge profits for micro-blogging website. Therefore, mining interest of microblog users and making good use of it is a very notable researcher direction.Taking sina weibo as the research object,a method of mining user interest in microblog based on behaviour is proposed in this paper, against the inaccurate problem of mining user interest in microblog site currently. First, the method analyzes the characteristics of microblog websites and the behaviour of microblog users, focuses on user behaviour of attention,releasing micro-blogs, comments and adding user tags, and selects the representative behavioral information of user interest by statistics and analysis. Afterwords, a user interest category classifier is constructed to mining interest from behaviour information which represents user interest. Fusing the mining results of different behaviour information with the harmonic factor, a method of mining user interest in microblog based on behaviour is proposed in this paper. Finally, a practical and feasible evaluation criteria and verification scheme is worked out. With the API offered by sina microblog open platform, experimental data is acquired. Compared with the actual user interest model, experiments are carried out against the feasibility, precision and sensitivity of the method of microblog user interest mining. Experimental results show that the method of user interest mining in microblog based on behaviour is feasible and that the mining result is more accurate. At the same time,the method has a more sensitive response against the drift of user interst.