Research On Credible User's Virtual Identity Association Of Social Network Platform

As social networking and social life become more and more integrated,more and more people have virtual identities on multiple platforms.Users may also have registered QQ and WeChat,Guest live and Betta live,Sina Weibo and Tencent Weibo.The abroad users may registered Twitter and Facebook.However,social networking platform generally adopted anonymous registration mechanism,for the false information and rumors left a loophole,the spread of potential safety problem is unable to avoid.The study of virtual identity was born in this situation.Based on the behavior characteristics,this paper studies the virtual identity association via hierarchical approach based on the credibility pre-assessment,the focus includes three areas:1.Propose a new method of combining the trustworthiness assessment and the virtual identity association.The PAC algorithm is designed to evaluate the credibility of the credit.The algorithm takes into account the factors such as the user's text,social relations and influence,and constructed a CVIA model.2.Compared with the previous work,the article takes the user name,registration information,social relations,text theme,behavior characteristics and other aspects into a comprehensive consideration,to build a hierarchical model,and strive to make the results more scientific and stable,reliable.3.The improved vector space model is designed as the basic method of similarity calculation.The FAHP algorithm is applied to the determination of the weight coefficient of the hierarchical model.The HVIA model is constructed and implemented.Finally,the HVIA algorithm is proved by the data set of the social network platform to be consistency,and the feasibility of constructing the CVIA model is proved by combining with the credibility pre-evaluation.And the model is analyzed and evaluated by calculating the precision,recall and F-measure value of identity association.In the best parameter configuration,the accuracy rate of the results can reach more than 90%,recall rate of more than 80%,F-measure value of more than 85%.The experimental results show that the performance of the model is good.