A fact-alignment LLM method detects clinical summarization hallucinations better than existing metrics, with correlations of 0.43 on controlled data and 0.37 on natural errors.
This approach makes it tractable to generate domain specific evaluation data when nat- ural hallucinations are difficult to find in large numbers
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Fact-Controlled Diagnosis of Hallucinations in Medical Text Summarization
A fact-alignment LLM method detects clinical summarization hallucinations better than existing metrics, with correlations of 0.43 on controlled data and 0.37 on natural errors.