Artículo: Towards Automated Generation of Object-Centric Event Logs to Support Software Development Workflow Analysis
Archivos
Fecha
Editor
Publicado en
Licencia Creative Commons
Resumen
Collaborative software development platforms such as GitHub provide a rich empirical basis for studying how collaborative and iterative development processes are enacted in practice. Repositories hosted on these platforms inherently involve multiple interconnected entities, including commits, issues, pull requests, user contributors, etc., that evolve concurrently and mutually influence one another. Although process mining techniques have been applied to the analysis of deployment pipelines and agile software processes, prior work has largely relied on case-centric perspectives, which oversimplify these complex interactions by forcing repository activity into a single-case notion. Object-centric event logs address these limitations by explicitly capturing event–object and object–object relationships, enabling a more faithful representation of collaborative and iterative development processes and supporting more comprehensive analyses of repository dynamics. However, generating object-centric event logs is non-trivial, especially from real-world scenarios, due to the complexity of mapping many-to-many relationships between entities. In this paper, we aim to enhance the analytical capabilities for software development workflows in two ways. First, we propose GitHub2OCEL, an automated mechanism for generating OCEL 2.0-compliant event logs from GitHub repository data. Second, we discuss the analytical potential of the generated logs by defining five analytical dimensions of workflow analysis, and exploring mechanisms to support advanced analysis and improve usability. The approach was validated on a real-world repository.


