TCMaster improves confidence-aware queries on Chinese medicine knowledge graphs
TCMaster: Confidence-Aware Querying and Workload-Guided Physical Design for Multi-Source Traditional Chinese Medicine Knowledge Graphs
Artificial Intelligence
Summary
Many databases about traditional Chinese medicine come from different sources with varying reliability. The authors created TCMaster, a tool that helps users search these databases more accurately by considering how trustworthy each piece of information is. It speeds up queries and filters out less reliable paths in the data, making answers to questions about medicine better by 20%. TCMaster combines many types of medical data into one network and optimizes how the data is stored and accessed.
What this means in practice
- •For pharmaceutical data teams: Use confidence-aware queries to improve retrieval accuracy in integrated drug and herb knowledge systems.
- •For medical informatics engineers: Build optimized knowledge graph systems for traditional medicine by leveraging workload-guided physical design.
Authors
Zheng Chen, Yuzhu Li, Haoxuan Li, Zhongde Zhang, Lianshun Jin, Peiwu Qin
Abstract
Multi-source knowledge graphs (KGs) need query mechanisms that expose reliability and exploit domain structure. This paper presents TCMaster, a property-graph query substrate for confidence-aware traversal and workload-guided physical design over Traditional Chinese Medicine KGs. TCMaster integrates pharmacopoeias, prescriptions, molecular databases, and LLM-extracted micro-semantics into a KG with approximately 221K entities and 723K base edges. It annotates edges with provenance-level confidence, rewrites Cypher queries with confidence predicates, ranks multi-hop paths under PRODUCT, MIN, or weighted-average policies, and uses ontology skew through direction selection, herb-attribute bitmaps, and materialized shortcut edges. On Neo4j, direction selection improves attribute lookup by a factor of 1.47, shortcuts accelerate high-fanout target counting by a factor of 4.42, confidence filtering removes 39.3 percent of low-quality heterogeneous paths, and KG retrieval improves TCMbench QA accuracy by 20.0 percentage points.