Softverse: Auto-Compute Citations to Software from Replication Files (github.com)

Softverse has launched an innovative tool that automatically computes citations for software used in social science research, leveraging replication files from 34 journals accessible on the Harvard Dataverse. This initiative is significant for the AI and machine learning community as it enhances the reproducibility of research by providing clear metrics on software usage, enabling researchers to better understand the tools that contribute to their findings. The technical framework of Softverse includes a series of commands that allow users to collect datasets, analyze imports from scripts, and generate citation metrics effortlessly. Users can install it via Python's package manager under version 3.11 or higher, utilizing various commands to interact with sources like Dataverse, Zenodo, and ICPSR. The tool streamlines the process of gathering data, searching for relevant scripts, and synthesizing analyses, thus supporting reproducibility and transparency in research. By tallying software library usage through regex file analysis, Softverse not only creates a reference for citations but also highlights the most commonly used packages, fostering deeper insights into research methodologies across disciplines.