Abstract
Enterprise data management (EDM) is a critical success factor for leveraging the increasing data volume and variety but relies mostly on manual efforts and human intervention. While first studies show the significant potential of ML techniques, they have a strong technical focus and only address isolated problems. Against this backdrop, our study sheds light on how ML techniques can advance EDM. Based on the analysis of 43 ML cases, our study makes two main contributions: First, we suggest a taxonomy that links ML applications to concepts from EDM and data curation. Second, we identify nine archetypes that provide an overview of typical application areas of ML in EDM. We find that ML techniques induce a shift from manual data maintenance in a reactive mode to data creation in a proactive mode. Our analysis also reveals that some archetypes build on a rich body of research from the database community.
Recommended Citation
Fadler, Martin and Legner, Christine, "Understanding the Impact of Machine Learning on Enterprise Data Management: A Taxonomic Approach" (2019). Proceedings of the 2019 Pre-ICIS SIGDSA Symposium. 22.
https://aisel.aisnet.org/sigdsa2019/22