Anchor Link Prediction Based on Trusted Anchor Re-identification

Dongwei Zhu, Yongxiu Xu, Lei Zhang, Minghao Tang, Wenhao Zhu, Hongbo Xu

Abstract: Cross-social network anchor link prediction plays a pivotal role in downstream tasks, such as comprehensively portraying user characteristics, user friend recommendations, and online public opinion analysis, which aims to find accounts that belong to the same natural person on different social networks. It is a common method to use manually marked anchors or anchors inferred through autonomous learning as supervisory information to guide the prediction of subsequent anchor links. However, the credibility of the anchor is not discussed. In this paper, to address this problem, we propose a new framework that can simultaneously complete the identification of trusted anchors and the prediction of anchor links across social networks under a unified framework. The proposed method can effectively identify non-trusted anchor links and improve the accuracy of the anchor link prediction model through the reconstruction of trusted anchors. Extensive experiments have been conducted on two large-scale real-life social networks. The experimental results demonstrate that the proposed method outperforms the state-of-the-art models with a big margin.