Link prediction via controlling the leading eigenvector

Link prediction is a fundamental challenge in network science. Among various methods, similarity-based algorithms are popular for their simplicity, interpretability, high efficiency and good performance. In this paper, we show that the most elementary local similarity index Common Neighbor (CN) can be linearly decomposed by eigenvectors of adjacency matrix target network, with each eigenvector’s contribution being proportional to square corresponding eigenvalue. As many real networks, there huge gap between largest eigenvalue second eigenvalue, CN thus dominated leading eigenvector much useful information contained other may overlooked. Accordingly, propose parameter-free algorithm ensures contributions secondary same. Extensive experiments on networks demonstrate performance proposed remarkably better than well-performed indices literature. A further adjust shows superiority over state-of-the-art tunable parameters its competitive accuracy lower computational complexity.