DSP: Deep Sign Prediction in Signed Social Networks

In a signed social network, users can express emotional tendencies such as: like/dislike, friend/foe, support/oppose, trust/distrust to others. Sign prediction, which aims to predict the sign of an edge, is an important task of signed social networks. In this paper, we attempt to tackle the problem of sign prediction by proposing a Deep Sign Prediction (DSP) method, which uses deep learning technology to capture the structure information of the signed social networks. DSP considers the "triangle" structures each edge involves comprehensively, and takes both the "balance" theory and the "status" theory into account. We conduct experiments and evaluations on five real signed social networks and compare the proposed DSP method with multiple state-of-the-art methods. The experimental results show that the proposed DSP method is very effective and outperforms other methods in terms of four metrics (AUC, binary-F1, micro-F1, macro-F1).