A Two-Stagse Approach for Social Identity Linkage Based on an Enhanced Weighted Graph Model

Abstract

Social identity linkage refers to identify the accounts belong to the same person across different social networks. This work can assist in building more complete social profiles, which is valuable for many social-powered applications. In this paper, we propose a two-stage approach to improve the efficiency and accuracy of large-scale social identity linkage. The first stage deals with the seed set enrichment problem and focuses on exploring a larger set of seeds with greater precision. The second stage deals with the global propagation problem and focuses on finding more matched pairs with lower computation. Moreover, we propose an enhanced weighted graph model to deeply investigate the structural characteristics. We also develop an attribute representation method to reduce the impact of missing attributes. Finally, we evaluate our method based on the datasets collected from two popular social networks in China. And the experimental results demonstrate that our method outperforms other state of the art algorithms.