Detecting Circles on Ego Network Based on Structure

As the online social network (OSN) develops, the social media data can be easily obtained, which leads to a large scale social network data. In our offline social networks, social relationships that an individual (ego) maintains with other people (alters) can also be organized into circles or groups according to the ego network model [1]. As to online social network, in order to help users cope with personal social network, many OSN sites provide services for users to group their friends. With many restrictions, such as on Facebook one cannot access other´s personal information unless they establish ties with each other, the network structure is the only information known by users. In this paper, we propose an algorithm to find circles on personal social network mainly from the structural view. Based on the defined structural similarity and the condition where nodes constitute a circle, the algorithm is able to cluster the most similar nodes into the same circle. We then give proof to demonstrate the convergence of our algorithm. To show the effectiveness of our algorithm, we realize the algorithm on Facebook datasets. Compared with the community detection algorithm, the results are evaluated from the visualized quantified aspect. From the results, we find that community detection algorithm cannot directly deal with the circle detection problem due to the characteristic of ego network and our algorithm can effectively handle the circle detection on ego network.