基于CLA算法的跨社交平台用户身份匹配

近几年,面向跨社交平台识别分布在不同社交网络上的同一用户依然是一个未解决的难题。该研究可以解决商业应用、资源整合、好友推荐等方面的相关问题。现有的算法如通过文本挖掘、单纯的用户属性无法取得良好的效果。提出CLA(Combined Link and Attribute)算法实现用户身份匹配。通过好友亲密度获得候选用户,结合基于网络结构的链接信息和用户属性信息进行用户匹配度计算。其中,链接信息相似度利用朋友匹配度计算得到。将该算法应用于多种社交网络,实验结果表明,该算法效果优越于传统的算法效果。

CROSS-SOCIAL PLATFORM USER IDENTIFICATION BASED ON CLA ALGORITHM

In recent years, identifying the same users distributed on different social networks for cross-social platforms is still an unsolved problem. This research can solve the problems related to business applications, resource integration, friend recommendation and so on. Existing algorithms such as text mining and simple user attributes cannot achieve effective results. This paper proposed a combined link and attribute ( CLA ) algorithm for user identification. The candidate users were obtained through calculating friend intensity (FI), and then the user matching degree was calculated by combining the link information based on network structure and user attribute information. The similarity of link information was calculated by the friend matching degree. The algorithm was applied to a variety of social networks. The experimental results show that the CLA algorithm is superior to the traditional algorithm.