LLMs reach above 0.8 accuracy on coarse identification of highly recommended articles but drop in fine-grained rating and commenting tasks, with supervised fine-tuning performing best and moderate positive correlations to citations.
You assess the quality, relevance, and potential impact of the research and provide your expert opinion on its importance
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Large language models for post-publication research evaluation: Evidence from expert recommendations and citation indicators
LLMs reach above 0.8 accuracy on coarse identification of highly recommended articles but drop in fine-grained rating and commenting tasks, with supervised fine-tuning performing best and moderate positive correlations to citations.