Enhancing Knowledge Preservation in Machine Learning Research: Jupyter Notebooks as an Interactive Documentation Tool

Tobias Schmieg, Fabian Schoch, Nicholas Schloer, Carsten Lanquillon, Carsten Wittenberg, Nicolaj C. Stache

Abstract: Jupyter Notebooks have gained widespread popularity in research and education due to their ability to integrate code, explanatory text, and visualizations within a single document. This study examines their potential as a medium for preserving and transmitting domain-specific knowledge in machine learning research. A significant challenge encountered in research institutions pertains to the loss of specialized knowledge when projects reach their conclusion or researchers transition to new roles. Traditional documentation practices often fail to capture the depth of methodologies, design decisions, and experimental workflows, making it difficult for new researchers to build upon prior work.

External IDs:dblp:conf/hci/SchmiegSSLWS25