Researchers have introduced RWGBench, a new benchmark designed to evaluate the quality of related work generation (RWG) in large language models. Unlike existing metrics that focus on text similarity, RWGBench assesses RWG based on citation decision-making, which is crucial for scholarly writing. The benchmark, constructed from a large corpus of computer science papers, evaluates citation selection, contextual appropriateness, organization, and discourse. This approach reveals limitations in current systems that are obscured by traditional evaluations and aligns better with expert judgment. AI
IMPACT Provides a more accurate evaluation of AI's ability to assist in scholarly writing, potentially improving research tools.
RANK_REASON The cluster describes a new academic benchmark for evaluating AI systems. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- Anzhe Xie
- arXiv
- CatalyzeX Code Finder for Papers
- Connected Papers
- DagsHub
- Gotit.pub
- Hugging Face
- Litmaps
- RWGBench
- ScienceCast
- scite Smart Citations
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
