CRISP: Characterizing Relative Impact of Scholarly Publications
📰 ArXiv cs.AI
CRISP uses large language models to jointly rank cited papers within a citing paper, enabling relative comparisons of their impact
Action Steps
- Train a large language model on a dataset of scholarly publications
- Use the trained model to jointly rank all cited papers within a citing paper
- Mitigate positional bias by ranking each list multiple times in a randomized order
- Analyze the ranked lists to determine the relative impact of each cited paper
Who Needs to Know This
Researchers and academics in AI and related fields can benefit from CRISP to better understand the impact of scholarly publications, and developers of AI models can apply this method to improve their models' ability to analyze citation context
Key Insight
💡 Jointly ranking cited papers within a citing paper enables relative comparisons of their impact, providing a more comprehensive understanding of a paper's influence
Share This
📚💡 CRISP: a new method for characterizing the relative impact of scholarly publications using large language models #AI #research
DeepCamp AI