Analyze Smarter, Not Harder: Generating Publication-Ready Citation Diversity Reports with LLMs

Session Title for Website and App: Analyze Smarter, Not Harder: Generating Publication-Ready Citation Diversity Reports with LLMs

Special Session Program

Description: Recent work has highlighted a citation imbalance across several fields of science, including biomedical engineering, in which women and minority authors are under-cited relative to their male and majority colleagues [1,2]. This systemic phenomenon not only negatively impacts the visibility and career success of minority scientists but also biases the prominent research questions of a field. One proposed solution is the inclusion of a Citation Diversity Report/Statement in a manuscript immediately preceding its reference section. To prepare these, manuscript authors compile the gender and race/ethnic demographic information of cited authors, subsequently enabling reflection over their citation diversity and offering an opportunity to correct any citation blind spots before publication. Current methods of sourcing gender and race/ethnic demographic information can be laborious, discouraging researchers from investing time into developing these reports. However, after testing 27 different LLMs from 7 companies, we have discovered that certain large language models (LLMs) can be used as all-in-one tools to easily generate publication-ready citation diversity reports within minutes. In this pre-conference workshop, audience members will be guided through the process of prompting the adept LLMs to craft their own citation diversity reports for their in-progress manuscripts. In the second half of the workshop, we will hold a panelist Q&A on LLM practices in publishing and efforts to diversify science.