Simple Augmentations of Logical Rules for Neuro-Symbolic Knowledge Graph Completion

TL;DR

The paper tackles the challenge of limited rule coverage in neuro-symbolic KG completion by introducing three lightweight augmentations: abductive rule transformation, rule inversion with relation inverses, and PCA-guided random-walk rule discovery, complemented by pruning. These augmentations are model-agnostic and applied across multiple NS-KGC baselines, yielding consistent performance gains and, in some cases, state-of-the-art results for NS-KGC on datasets like WN18RR; they also demonstrate substantial improvements when combined with ExpressGNN. The work further analyzes rule quality and trade-offs, showing that augmentation-based gains largely stem from increased high-quality rules rather than simply adding more rules, and provides qualitative evidence of interpretable augmented rules. Overall, the findings suggest that simple, principled rule augmentations can significantly enhance symbolic inferences in NS-KGC and should be integrated as a standard step in rule-based KG completion pipelines.

Abstract

High-quality and high-coverage rule sets are imperative to the success of Neuro-Symbolic Knowledge Graph Completion (NS-KGC) models, because they form the basis of all symbolic inferences. Recent literature builds neural models for generating rule sets, however, preliminary experiments show that they struggle with maintaining high coverage. In this work, we suggest three simple augmentations to existing rule sets: (1) transforming rules to their abductive forms, (2) generating equivalent rules that use inverse forms of constituent relations and (3) random walks that propose new rules. Finally, we prune potentially low quality rules. Experiments over four datasets and five ruleset-baseline settings suggest that these simple augmentations consistently improve results, and obtain up to 7.1 pt MRR and 8.5 pt Hits@1 gains over using rules without augmentations.