π€ AI Summary
In knowledge graph reasoning, conventional chain-based rules suffer from limited expressiveness and spurious triggering, leading to erroneous inferences. To address this, we propose βtree-shaped rulesββa novel paradigm that transcends linear rule structures by defining semantically richer and logically more rigorous rule forms. We design a plug-and-play rule dimensionality-upgrading framework integrating rule mining, subgraph pattern matching, and logical structure optimization, enhanced by heuristic pruning and confidence-driven tree generation. Evaluated on link prediction across four public benchmarks, tree-shaped rules consistently outperform chain-based baselines, achieving average MRR improvements of 5.2%β9.7%. The approach simultaneously enhances both interpretability and inference accuracy. Our code and datasets are publicly available.
π Abstract
With good explanatory power and controllability, rule-based methods play an important role in many tasks such as knowledge reasoning and decision support. However, existing studies primarily focused on learning chain-like rules, which limit their semantic expressions and accurate prediction abilities. As a result, chain-like rules usually fire on the incorrect grounding values, producing inaccurate or even erroneous reasoning results. In this paper, we propose the concept of tree-like rules on knowledge graphs to expand the application scope and improve the reasoning ability of rule-based methods. Meanwhile, we propose an effective framework for refining chain-like rules into tree-like rules. Experimental comparisons on four public datasets show that the proposed framework can easily adapt to other chain-like rule induction methods and the refined tree-like rules consistently achieve better performances than chain-like rules on link prediction. The data and code of this paper can be available at https://anonymous.4open.science/r/tree-rule-E3CD/.