Social Circles Discovery Based on Structural and Attribute Similarities

摘要

Due to huge amount of information which is generated on online social networks every day, how to organize their personal social networks to cope with the overwhelming information effectively has become a difficult problem. In this paper, a new modified method based on Clique Percolation Method (CPM) algorithm is proposed to be used to discover users' social circles automatically, in which, there are two improvements by integrating the structural and attribute similarities into a unified framework. Experiments on several real datasets demonstrate that the result of social circles identification used by the modified method proposed in this paper is better compared with CPM method.