Query-Conditioned Knowledge Alignment for Reliable Cross-System Medical Reasoning

2026-05-18

AI Summary

This work addresses the limitations of traditional cross-system medical knowledge alignment methods, which treat entity alignment as a static matching task and overlook query context, non-bijective relationships, and directional sensitivity—hindering reliable integrative reasoning. To overcome these challenges, the authors propose the Query-Conditioned Entity Alignment (QCEA) framework, which reformulates alignment as a query-driven, dynamic many-to-many correspondence problem. Specifically, QCEA uses the source entity description as a query to rank candidate entities in the target knowledge graph, enabling context-aware alignment. The approach integrates semantic encoding, graph representation learning, and direction-aware transformation, and leverages retrieval-augmented generation (RAG) for validation. Evaluated on the TCM-WM dataset, QCEA substantially outperforms baseline methods, achieving notable gains in Hit@K and MRR metrics, and significantly improves evidence retrieval accuracy and answer reliability in downstream RAG tasks.

Abstract

Cross-domain knowledge alignment is essential for integrating heterogeneous medical systems, yet existing approaches typically treat entity alignment as a static matching problem, ignoring query context and cross-system asymmetry. This limitation is particularly critical in integrative medical settings, where correspondence between concepts is inherently context-dependent, non-bijective, and direction-sensitive. In this paper, we propose Query-Conditioned Entity Alignment (QCEA), which reformulates entity alignment as a query-conditioned correspondence problem. Instead of learning a fixed mapping between entity representations, QCEA treats the textual description of a source entity as a query and ranks candidate entities in the target graph, enabling context-dependent alignment. The framework integrates semantic encoding, graph-based representation learning, and a direction-aware transformation module to capture asymmetric and many-to-many correspondence across heterogeneous knowledge systems. We evaluate QCEA on TCM--WM knowledge graphs derived from SymMap, covering both symptom alignment and herb--molecule alignment tasks. Experimental results show consistent improvements over representative baselines, particularly on rank-sensitive metrics such as Hit@K and MRR. Furthermore, downstream retrieval-augmented generation (RAG) experiments demonstrate that improved alignment leads to better evidence retrieval, stronger grounding, and higher answer accuracy. These findings highlight that alignment is not merely a data integration step, but a key factor that shapes knowledge accessibility and reliability in cross-system medical reasoning.

Problem

• cross-domain knowledge alignment
• entity alignment
• query context
• asymmetric correspondence
• heterogeneous medical systems

Innovation

• Query-Conditioned Entity Alignment
• Cross-System Medical Reasoning
• Asymmetric Knowledge Alignment
• Heterogeneous Knowledge Graphs
• Retrieval-Augmented Generation