Link Prediction Research Based on Visual Analysis: Expansion of the LERTR Index and System Validation

Link prediction is widely used in areas like social networks, bioinformatics, recommendation systems, IoT, and healthcare to uncover latent information in complex networks. While similarity-based methods are simple and interpretable, they often overlook high-order structures and additional attributes, limiting their performance. To address this, we developed LPExplorer, a system that integrates the LERTR index (combining RA and LCP principles) for interpretable and accurate predictions across three-hop paths. LPExplorer visually represents network structures, resource flows, and parameter impacts, allowing users to sort and filter predictions effectively. Experimental results demonstrate its strong performance in analyzing and exploring complex networks.