Self-Adaptive Optimized Link Prediction Based on Weak Ties Theory in Unweighted Network

Recently, link prediction becomes the focus of researchers from various fields of science, and considerable achievements have been obtained. Algorithms based on similarities constitute the research mainstream. This paper proposes an optimization to the classical algorithms: Adamic-Adar (AA) and Resource Allocation (RA) with the same foundation in local similarity. Herein, weak ties theory will be introduced to improve accuracy of AA and RA. Although Lü et al. discussed the functions of weak ties in AA and RA, only weighted network was considered. Then this paper will pay more attention to unweighted and undirected simple network. Theoretical analysis will be processed in advance, and at the end, experiments will be performed in five real networks to compare classical algorithms with the optimized index in the network through numerical analysis.