Identifying Users' Interest Similarity Based on Clustering Hot Vertices in Social Networks

Abstract

Identifying users' similarity is a very important researching point because its result can be applied to many application systems. In social networks, the user circles are built not only based on their relationships in real-life, but also on common interests. Some existing approaches cannot fully capture users' similarity from the perspective of their common interests, while some other approaches are too time-consuming or space-consuming. In this paper, we propose a method of identifying users' interest similarity based on clustering Hot Vertices (HotV). A hot vertex in a social network is an account which has a large number of fans. The approach extracts users' common interests by mining and clustering the hot vertices that the two users are following simultaneously. Both the experiment and theoretical analysis have proved that the proposed approach makes a significant improvement on the precision of similarity measuring with a relatively low time and space complexity.