The Scientific Research Taste Benchmark (SciTaste-Bench): A Multi-Disciplinary Benchmark for Assessing Large Language Models' Judgment of Research Taste in Scientific Papers
cs.DL
Submitted: 2026-10-09
Updated: 2026-10-09
Terminology
Related papers
- Persistence Paradox in Dynamic Science: Evidence from the Deep Learning Revolution
- How Deep Do Large Language Models Internalize Scientific Literature and Citation Practices?
- MTRACE: Multilingual Retrieval-Augmented Generation for Temporally Diverse Text Corpora
- Eleven Years of BRACIS: A Meta-Scientific Study of the Brazilian Conference on Intelligent Systems
- Evidence-Based Scientific Question Discovery: A Framework with Historical Backtesting