LINK PREDICTION IN SIGNED SOCIAL NETWORKS BASED ON NEUTRAL RELATIONSHIP

With the development of social networking technology, the problem of link prediction in signed social networks has a wide range of applications and also poses great challenges. The related works are mainly focused on social theories, but the impact of neutral social relationships common in real social networks on prediction has not been fully studied. Based on the exploration of neutral relationships, the SILENT algorithm is proposed to try to study the above problem from the following perspectives: 1) Unified prediction of various possible social relationships between node pairs; 2) Exploring the influence of neutral relationships on the prediction. Evaluation experiments show that SILENT can effectively predict complex social relationships between node pairs, and outperforms the state-of-the-art algorithms in terms of performance.