An improved density peaks-based clustering method for social circle discovery in social networks

With the development of Internet, social networks have become important platforms which allow users to follow streams of posts generated by their friends and acquaintances. Through mining a collection of nodes with similarities, community detection can make us understand the characteristics of complex network deeply. Therefore, community detection has attracted increasing attention in recent years. Since the problem of discovering social circles is posed as a community detecting problem, hence, in this paper, targeted at on-line social networks, we investigate how to exploit user's profile and topological structure information in social circle discovery. Firstly, according to directionality of linkages, we put forward in-link Salton metric and out-link Salton metric to measure user's topological structure. Then we propose an improved density peaks-based clustering method and deploy it to discover social circles with overlap on account of user's profile- and topological structure-based features. Experiments on real-world dataset demonstrate the effectiveness of the proposed framework. Further experiments are conducted to understand the importance of different parameters and different features in social circle discovery.