Understanding Open-Source NLP Artifact Adoption Through Information Systems Success Factors

Abstract

Open-Source Software (OSS) movement has significantly shaped the landscape of data science, particularly in the subfield of natural language process (NLP). Despite the popularity and rapid growth of OSS tools in the field of data science, prior IS literature did not examine the adoption of NLP artifacts through the lens of OSS success. This study applies and extends the DeLone and McLean Information Systems Success Model to the context of OSS in the domain of NLP. Our extended model examines the moderating effects of task type on the relationship between system quality, information quality, and adoption. In this study, we gather model cards of NLP artifacts, and their download/endorsement counts from Hugging Face and empirically examine the adoption behavior of OSS NLP artifacts. Our expected findings would suggest that system quality affects adoption more for analysis tasks compared to generation tasks, while information quality affects adoption more for generation tasks compared to analysis tasks.