Link prediction on signed social networks based on latent space mapping

Link prediction is an essential research area in social network analysis. In recent years, link prediction in signed networks has drawn much concentration of the researchers. To predict potential positive and negative links, we should predict not only the existence of the link between the nodes, but also the sign and the probability of the existence of the link. In addition, the link prediction result should satisfy the social balance and status theories as much as possible. In this paper, we propose an algorithm for link prediction in signed networks based on latent space mapping. Taking the social balance and status theories into consideration, we define a balance/status coefficient matrix to reflect the balance/status constrains on the signs of the unknown links. We also present the concept of signed degree ratio and the signed degree ratio-based similarity between the node pairs to measure probability of the signed links. We propose a latent space-based model for the connections in a signed network which combines the topological structure and the balance/status constrains. An alternative iteration algorithm is proposed to optimize the model, and its convergence and correctness are proved. By this alternative iteration method, time complexity of our algorithm is reduced greatly. Empirical results on real world signed networks demonstrate that the algorithm proposed can achieve higher quality predicting results than other algorithms.