User Personality And Behavior Analysis Based On Social Networking Services

In recent years social networks have sprung up, there are some well-known social networks websites such as sina microblogging, QQ space in china, and Facebook, Twitter in foreign. Social networks is changing people’s life-style and social way. People gradually accepted and used to sending photos, writing the blogs, updating status on social networks. On the other hand, the user’s performance on social networks also tend to differentiate, for example, some people just like to browse without posting content, while others prefer to blog but rarely send photos. These different acts are not haphazard, but contain some certain rules. How to effectively analyze the user’s behavior, dig deeper law behind the behavior and thus to provide a personalized service is becoming a major difficulty. Currently, analysis on user behavior mainly focused on user’s behavior data, not made use of user’s published text, such as the user’s status and blogs. On the other hand, the analysis is not involved with user’s personality model. If finding the intrinsic link between a user’s personality and behavior, it must be able to provide new theoretical support for the analysis of social networks.The main work of this paper includes the following aspects:1. Determine the analytical method. Firstly, this paper discusses the current development of domestic social networks and access to social network user’s data, then selects RenRen as research object. The data collecting is implemented by building a in-site application, and the application offers some personality tests to users.2. Build the in-site app. The personality test selects Big Five personality test scales. The average score for each character ingredients into five grades by the using of the normal distribution, feedback to the user is based on the grades. Then this paper adopts Flex to implement UI, Java to implement the web server, MySQL database to store user’s data. The application gets the user’s authorization by OAuth authentication, reads and stores user’s personal information and UGC data through API.3. Process uses’data. The application records not only the user’s personal information but also UGC data. Based on these data, this paper calculates the user behavior statistics, including user’s frequency of posting status, blogs, or sharing a blog, albums and so on. In addition, this paper applys some semantic analysis on user’s UGC content, including text segmentation and statisting word frequency, the adjustment of different words’weight. Finally this paper uses principal Component analysis to simplify the data obtained.Based on the above behavior data and semantic data, this paper uses linear regression and decision tree algorithm to predict the user’s gender, age, personality. At last the predicted results are compared with known records, to verify the effectiveness of the algorithm.