Weibo User Attributes Certification Research And Application

With the rapid development of Internet,it has penetrated into people’s daily life.As a social networking platform,microblogging is popular in society and has quickly become a prevailing fashion for chatting and information acquisition.By virtue of hundreds of millions of users,the major microblogging platforms possess huge amounts of user information,beneath which inestimable commercial value lies.It is of great importance how to correctly apply the data to discover underlying crucial knowledge so as to better understand users’ behavior and achieve substantial value.Under such motivations,this thesis focuses on attribute certification for Weibo users.Several important properties authentication algorithms for Weibo users are studied.A new professional property authentication algorithm for Weibo users based on word vector distance is proposed,which predicts user’s occupation by measuring the distance between the released blogs content by user and professional words.Moreover,we use the Word2vec,which is a word-vector conversion tool based on neural network,to improve the algorithm’s accuracy.The experiments based on real user data show that the algorithm’s accuracy can be as high as nearly 80%.At the same time,we also study the algorithm for user’s another social attribute,which is user role.A comprehensive evaluation index named U-Score for user role analysis is proposed,which involves many different types of hierarchical indexes and takes five different factors into account:influence,activeness,centricity,credibility and importance.And the algorithm uses analytic hierarchy process to calculate the weight of different characteristics.The experimental results show that this method is feasible for Weibo user role analysis in quantifying various indicators of the user.This thesis also studies on the Weibo user’s gender attribute certification.According to analysis on characteristics of user data gathered from Weibo platform,three different types of user characteristics are integrated to classify gender attribute.The classification accuracy can be over 90%.At the same time,a Weibo user attributes certification system is developed by using the above attributes authentication algorithms.The system includes three modules:data acquisition,data storage and data mining.In the module of data mining authentication,the system integrates the above three kinds of user attributes authentication algorithms,and accomplish user attributes certification.