新浪微博影响力分析汇报.ppt

《新浪微博影响力分析汇报.ppt》由会员分享，可在线阅读，更多相关《新浪微博影响力分析汇报.ppt（18页珍藏版）》请在淘文阁 - 分享文档赚钱的网站上搜索。

1、User Influence Analysis on Micro BlogCentral China Normal UniversityYong Zhang,Jiaqin Mo,Tingting He2012.11.01OutlineBackgroundRelated WorkBasic AnalysisMethodExperimentConclusionBackgroundlSina Weibo one of the most notable and widely used micro-blogging serviceslInfluence the potential of an actio

2、n of a user to initiate a further action by another userlInfluence Users make trends or innovations be adopted faster,help decision markers develop more effective strategieslSignificance one may achieve a large-scale chain-reaction of influence driven by word-of-mouth with a very small marketing cos

3、tBasic AnalysisFollower number distribution Number of followers.average follower number:3384270.7%of the users are above the averageBasic AnalysisRelation between Weibo number and follower numberWeibo number and follower number.not a positive correlationfollower number is not the key factor for user

4、s to post WeiboBasic AnalysisUser interaction analyzingFigure.3 Mention number and retweet number.half of the Weibos contain action many users tend to interact with others when they post Weibo,interaction is a very important action on Sina WeiboRelated workMethods to identify user influence on micro

5、 bloggingIn-degreeRatio between in-degree and out-degreePageRankXinRankAction-based network modelMethodThree important actions on microbloglFollow Follow Network Matrix FlRetweet Retweet NetworkMatrix RlComment Comment NetworkMatrix CMethodlMatrix F lMatrix R lMatrix CMethodCombination matrix MExamp

6、le：AAABBB52followretweetcomment=1,=0.05,=0.05F(A,B)=1R(A,B)=2C(A,B)=5W(A,B)=1*1+2*0.05+5*0.05 =1.35MethodRank AlgorithmsPageRankWeighted PageRankHITS(Hyperlink-Induced Topic Search)ExperimentDataset(analysis)467823 follow links67832 followees400000 followers389092 users Experiment1142 users129088 retweet links 225213 comment links.ResultResultResultResultConclusionsIntroduced three basic“interpersonal”actions:follow,retweet and comment action.Propose a model and apply it to Weighted PageRank and HITS algorithms.Comprehensive and stable method