CiteFairly: Toward the improvement of gendered citation practices in the software engineering field

Several studies have established that citation biases are pervasive in several STEM disciplines (e.g., Neuroscience, Physics, Computer Science). Indeed, scientific articles written by authors belonging to certain socio-cultural groups (e.g., women, Hispanics, Black) are usually significantly less cited than articles written by authors belonging to dominant groups. This trend is more exacerbated in the software engineering (SE) field where articles written by researchers from some socio-cultural groups (e.g., women) are considerably under-cited. Since the number of citations plays a key role in scientific circles, unfairness in citation practices therefore has an adverse impact on the recruitment, salaries, retention, promotion, and recognition of some researchers. To address this issue, we propose a novel approach to improve citation diversity in the reference lists of manuscripts that will be submitted for publication in software engineering venues. Thus, for each manuscript under analysis, our approach first identifies potential citation biases in its reference list, then attempts to mitigate these biases by recommending to its authors relevant articles written by other authors from diverse backgrounds. We implemented our approach as an online tool called CiteFairly . The latter is built atop several bibliometric tools such as: OpenAlex, Gender-API , and Semantic Scholar . To validate our approach, we conducted various experiments on the reference lists of papers published in several editions of well-established Software Engineering conferences. Our experimental results show that our approach can accurately identify and mitigate gendered citation biases in the analyzed reference lists with automated gender distributions deviating from manual annotations by no more than 5 percentage points on average.