Sentiment Analysis of Citations in Scientific Articles Using ChatGPT: Identifying Potential Biases and Conflicts of Interest
Walid Hariri
TL;DR
The paper addresses the need to quantify the sentiment of citations to better understand scholarly impact and potential biases in the literature. It proposes using ChatGPT's transformer-based NLP to perform sentiment analysis on citations, detect biases and conflicts of interest, and assess their contexts within articles. The work outlines domain-aware keyword cues, analytical approaches for positive/negative citations, and applications in reviewing, research, and editorial decision-making, including preprint contexts. The findings suggest that AI-assisted citation sentiment analysis can improve objectivity, transparency, and efficiency in scholarly evaluation, while highlighting important limitations such as model biases and ethical considerations.
Abstract
Scientific articles play a crucial role in advancing knowledge and informing research directions. One key aspect of evaluating scientific articles is the analysis of citations, which provides insights into the impact and reception of the cited works. This article introduces the innovative use of large language models, particularly ChatGPT, for comprehensive sentiment analysis of citations within scientific articles. By leveraging advanced natural language processing (NLP) techniques, ChatGPT can discern the nuanced positivity or negativity of citations, offering insights into the reception and impact of cited works. Furthermore, ChatGPT's capabilities extend to detecting potential biases and conflicts of interest in citations, enhancing the objectivity and reliability of scientific literature evaluation. This study showcases the transformative potential of artificial intelligence (AI)-powered tools in enhancing citation analysis and promoting integrity in scholarly research.