A new benchmark and six-class taxonomy with a 13-axis divergence ontology evaluates whether models can recover contextual reasons for biomedical claim disagreements, released as an article-disjoint silver dataset of 11,865 pairs.
Fact or fiction: Verifying scientific claims
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
BioDivergence: A Benchmark and Evaluation Framework for Hidden Contextual Contradictions in Biomedical Abstracts
A new benchmark and six-class taxonomy with a 13-axis divergence ontology evaluates whether models can recover contextual reasons for biomedical claim disagreements, released as an article-disjoint silver dataset of 11,865 pairs.