Unifying Deductive and Abductive Reasoning in Knowledge Graphs with Masked Diffusion Model
Yisen Gao, Jiaxin Bai, Yi Huang, Xingcheng Fu, Qingyun Sun, Yangqiu Song
TL;DR
DARK tackles the challenge of unifying deductive and abductive reasoning on knowledge graphs by modeling the joint distribution of a query and its conclusion with a masked diffusion framework. It introduces a self-reflective denoising mechanism to refine abductive hypotheses through deductive validation and a logic-exploration reinforcement learning strategy to discover diverse, plausible logical relationships by jointly masking queries and conclusions. The approach yields state-of-the-art results on both abductive and deductive reasoning benchmarks, and its theoretical and empirical analyses highlight favorable convergence properties and a favorable complexity profile. By leveraging bidirectional diffusion and open-world reasoning, DARK offers a scalable, interpretable framework for complex KG reasoning with incomplete data, with broad implications for semantic web applications and scientific discovery.
Abstract
Deductive and abductive reasoning are two critical paradigms for analyzing knowledge graphs, enabling applications from financial query answering to scientific discovery. Deductive reasoning on knowledge graphs usually involves retrieving entities that satisfy a complex logical query, while abductive reasoning generates plausible logical hypotheses from observations. Despite their clear synergistic potential, where deduction can validate hypotheses and abduction can uncover deeper logical patterns, existing methods address them in isolation. To bridge this gap, we propose DARK, a unified framework for Deductive and Abductive Reasoning in Knowledge graphs. As a masked diffusion model capable of capturing the bidirectional relationship between queries and conclusions, DARK has two key innovations. First, to better leverage deduction for hypothesis refinement during abductive reasoning, we introduce a self-reflective denoising process that iteratively generates and validates candidate hypotheses against the observed conclusion. Second, to discover richer logical associations, we propose a logic-exploration reinforcement learning approach that simultaneously masks queries and conclusions, enabling the model to explore novel reasoning compositions. Extensive experiments on multiple benchmark knowledge graphs show that DARK achieves state-of-the-art performance on both deductive and abductive reasoning tasks, demonstrating the significant benefits of our unified approach.