AI-Powered Citation Classification: Understanding Contrasting and Supporting Citations in Scholarly Research

Abstract

Citations play a critical role in the assessment of scholarly research. This study employs scite.ai, an artificial intelligence-based citation tool, which classifies citations into three categories: supporting, contrasting, and mentioning. The objective of the study is to explore the relationship between these citation types—particularly supporting and contrasting citations—and total citation counts in journals published by Elsevier. Using the journal-level search feature provided by Scite, data were manually collected from 3,151 Elsevier journals. Key metrics analysed include the Unweighted Scite Index (USI), total citation count, and the distribution of supporting, contrasting, and mentioning citations. Descriptive statistics and correlation analyses were conducted to examine these relationships. The results reveal statistically significant and positively correlated associations (p < 0.01) between supporting, contrasting, and total citation counts, highlighting a meaningful linkage between citation polarity and overall citation impact in scholarly publishing.