Crystal: Characterizing Relative Impact of Scholarly Publications

TL;DR

Crystal introduces a novel method using large language models to jointly rank and assess the impact of all cited papers within a citing paper, improving accuracy and efficiency over previous approaches.

Contribution

It proposes a joint ranking approach with LLMs that mitigates positional bias, enhancing citation impact assessment accuracy and scalability.

Findings

Crystal outperforms previous impact classifiers by +9.5% accuracy.

It achieves +8.3% F1 score improvement over prior methods.

The approach is more efficient, requiring fewer LLM calls and enabling scalable analysis.

Abstract

Assessing a cited paper's impact is typically done by analyzing its citation context in isolation within the citing paper. While this focuses on the most directly relevant text, it prevents relative comparisons across all the works a paper cites. We propose Crystal, which instead jointly ranks all cited papers within a citing paper using large language models (LLMs). To mitigate LLMs' positional bias, we rank each list three times in a randomized order and aggregate the impact labels through majority voting. This joint approach leverages the full citation context, rather than evaluating citations independently, to more reliably distinguish impactful references. Crystal outperforms a prior state-of-the-art impact classifier by +9.5% accuracy and +8.3% F1 on a dataset of human-annotated citations. Crystal further gains efficiency through fewer LLM calls and performs competitively with an open-source model, enabling scalable, cost-effective citation impact analysis. We release our rankings, impact labels, and codebase to support future research.