Research And Implementation Of Interest Recognition Algorithm For Microblog Users

With the coming of Web2.0, micro-blog has taken place of the traditional media as the representative of emerging social media, which is the symbol of the internet for individuals. As a kind of sharing and exchange platform, micro-blog pays more attention to timeliness and convenience, in which individuals can share their ideal and receive news every moment. However, with the explosive growth of information, how to recognize the interest of micro-blog users from the miscellaneous data is a hotspot in academic circle at home and abroad. Based on the current mainstream technology and the analysis of the characteristics of Chinese micro-blog, we combine the text information, image information and social relation information to recognize the user’s interest. The main work in this paper is listed as follows:Firstly, a micro-blog user oriented spider is implemented. With the input of a user’s ID of SINA micro-blog, the outputs include its text content, image content and social related user’s IDs, which is utilized to extract the interest of a user.Secondly, a new method is proposed to recognize the user’s interest that fusing micro-blog post information and social relation information. As for the micro-blog posts, we conduct it based on classification technology from the perspective of text and image information respectively, and fuse the result by linear regression as the interest distribution extracted from micro-blog posts. And as for micro-blog social information, we extract the related user list from the micro-blog content considering of forwarding, mention, following and comment and count the interest distribution of user through the dictionary of authoritative user. And then, we research how to merge the interest distribution micro-blog content information and social relation information to obtain the users final interest distribution.Finally, the system of interest recognition has been realized. We construct the model of interest recognition by the method we proposed above, based on which, the data collected from SINA micro-blog users is utilized to display the interest distribution for users.