融合用户行为和内容的微博用户影响力方法

Microblog user influence algorithm based on user behavior and content

为了模拟信息在微博环境中的传播情况,根据微博用户行为(发布、关注、转发和评论等)和微博内容,提出一种融合用户行为和内容的微博用户影响力算法。通过对微博用户行为的分析得到行为因子数据,进而计算出用户影响力的权值。利用微博用户内容建立词共现矩阵,继而运用狄利克雷分配(LDA)模型进行潜在主题分布的识别,通过KL(Kullback Leibler)散度的方法得到用户之间的相似性,最后结合用户影响力权值得到用户的影响力。实验表明此算法较为有效。

In order to sim ulate the propagation o f info rm atio n in m icro blog e n v iro n m e n t,th is paper presented a m icro blog userin flu e n ce algo rithm based on user b e h a vio r(p o s t,fo llo w ,fo rw a rd and com m ent) and con tent. B ehavior factor data could be obtainedthrough the analysis o f the be havior o f m icro blog u se r,a n d then the user in flu e n ce w eigh t could be ca lcu la te d . T h is algorithm established the word co-occurrence m a trix using the m icro blog co n te n t,a n d then used L D A ( late nt D iric h le t a llo c a tio n ) toid e n tify the po tentia l top ic d is trib u tio n . I t a p plied K L ( K u llb a c k L e ib le r) divergence to obtain the s im ila rity between u se rs,f i na lly acquired the user in flu e n ce based on using in flu e n ce w e ig h t. E xperim ents show the effectiveness o f this a lg o rith m .