Graph Structure Rule Learning via Contrastive Learning

Knowledge graph reasoning has attracted extensive attention because of its ability to alleviate the inherent incompleteness of knowledge graphs. However, existing approaches still face significant limitations in both expressiveness and interpretability. Embedding-based approaches often focus on chain-based rules instead of graph-structured rules, while traditional rule-mining techniques rely heavily on the “enumeration–verification” paradigm, which is computationally expensive on large-scale graphs and struggles to capture deep semantic correlations among rules. To address these challenges, we propose CGRL (Contrastive Graph Rule Learning), a novel framework for learning graph-structured rules through contrastive learning. CGRL enhances rule learning through a threefold mechanism. First, it applies an adaptive clustering strategy to generate soft type labels for entities, introducing semantic priors that improve rule generalization. Second, it designs a multi-hop neighborhood sampling method to construct semantic subgraphs and to generate high-quality positive and negative rule instances. Third, it employs a Transformer-based encoder and a contrastive learning objective to map rule pairs into a unified embedding space, thereby jointly capturing structural dependencies and semantic consistency. Extensive experiments on six public knowledge graph completion benchmarks demonstrate that CGRL achieves substantial improvements over state-of-the-art baselines. On the more complex datasets, CGRL yields average gains of 8.74% in MRR and 16.27% in Hit@1, confirming its effectiveness in discovering high-quality graph-structured rules and significantly enhancing reasoning performance.