Biomedical question answering improved with relation-based reasoning on knowledge graphs
Beyond Reachability: A Relation-Semantic Reasoning Framework for Biomedical Multi-hop QA
Computation and Language
Summary
Answering complicated biomedical questions often means finding a path through a knowledge graph that shows how pieces of information connect. The authors identify a problem where correct answers are found but the evidence supporting them isn’t always strong or clear. They introduce ReSemReason, a new approach that better matches questions to valid paths by focusing on how relations fit together semantically. This helps rank correct answers higher and improves the quality of reasoning used to reach them.
What this means in practice
- •For hospital data teams: Improve biomedical QA systems to provide better-justified answers from complex medical knowledge graphs, enhancing clinical decision support.
- •For bioinformatics platform developers: Enhance multi-hop question answering capabilities over biomedical data by integrating relation-semantic reasoning for more accurate knowledge retrieval.
Authors
Ziyang Ling, Ziyi Chen, Ronald X. Xu, Mingzhai Sun
Abstract
Biomedical multi-hop reasoning over knowledge graphs requires identifying a semantically valid relation sequence that supports the answer. Existing methods can retrieve correct answers, yet often rank them poorly within a large pool of seemingly relevant candidates. Even when a correct answer is retrieved, its associated evidence and reasoning path may fail to support the relation composition required by the question. We term this disconnect between answer reachability and evidential validity the reachable-but-invalid problem. We present ReSemReason, which shares a question-induced distribution over typed, directed relation schemas between candidate discovery and evidence selection. Schema compatibility guides graph search; ReSemAttn then uses relation-conditioned key/value projections and an uncertainty-weighted path prior to attend to candidate-specific evidence. A separate absolute-support score complements normalized attention when all retrieved paths are weak. The reported experiments on BioHopR, PrimeKGQA, and MedReason show improvements in answer ranking and answer-set quality. Path-discrimination analyses examine the distinction between endpoint correctness and evidential support. We also specify matched controls for search, attention, and schema uncertainty.