[2603.28325] Building evidence-based knowledge graphs from full-text literature for disease-specific biomedical reasoning

Computer Science > Computational Engineering, Finance, and Science arXiv:2603.28325 (cs) [Submitted on 30 Mar 2026] Title:Building evidence-based knowledge graphs from full-text literature for disease-specific biomedical reasoning Authors:Chang Zong, Sicheng Lv, Si-tu Xue, Huilin Zheng, Jian Wan, Lei Zhang View a PDF of the paper titled Building evidence-based knowledge graphs from full-text literature for disease-specific biomedical reasoning, by Chang Zong and 5 other authors View PDF HTML (experimental) Abstract:Biomedical knowledge resources often either preserve evidence as unstructured text or compress it into flat triples that omit study design, provenance, and quantitative support. Here we present EvidenceNet, a framework and dataset for building disease-specific knowledge graphs from full-text biomedical literature. EvidenceNet uses a large language model (LLM)-assisted pipeline to extract experimentally grounded findings as structured evidence nodes, normalize biomedical entities, score evidence quality, and connect evidence records through typed semantic relations. We release two resources: EvidenceNet-HCC with 7,872 evidence records, 10,328 graph nodes, and 49,756 edges, and EvidenceNet-CRC with 6,622 records, 8,795 nodes, and 39,361 edges. Technical validation shows high component fidelity, including 98.3% field-level extraction accuracy, 100.0% high-confidence entity-link accuracy, 87.5% fusion integrity, and 90.0% semantic relation-type accuracy. In downstre...