Abstract This article presents a multidisciplinary dataset of research papers annotated with software mentions, built by the SoFAIR project in collaboration with the CLARIN-PL team. The SoFAIR project aims to support the reproducibility of research and the reuse of data and to recognise the contribution of software developers. The dataset was created through a workflow consisting of text selection, automatic preannotation of documents using the Softcite software-mentions tool, and manual gold standard annotation. The dataset contains more than 9,000 software mentions, divided into 10 categories, as well as more than 2,000 relationships between mentions, which come from almost 500 texts belonging to 18 scientific disciplines. The dataset presented here contains detailed data and statistics that can be used to evaluate language models or to further develop software extraction tools. The dataset is available in TEI XML format and has been made available under a CC-BY license.
Rosiński et al. (Thu,) studied this question.