History Matters: Enhancing JIRA Software Development Workflow Models with Process Mining
Year: 2026
Type: preprint
Abstract: Issue-tracking systems such as JIRA record rich event data, yet discovered workflow models rarely make use of the contextual and behavioural information in issue reports. We investigate how process mining can enhance software development workflow models using a public Red Hat JIRA event log of 63,993 events, 5,759 issues and 175 projects. We discover a baseline Inductive Miner model and then apply three enhancement techniques: trace filtering, trace clustering and decision mining. Trace filtering raises precision from 0.526 to 0.829 and generalisation from 0.853 to 0.977 while keeping fitness at 0.995. Trace clustering raises weighted precision to 0.836 but lowers generalisation to 0.814. Decision mining raises routing accuracy from 0.701 to 0.762 and reduces routing uncertainty by 16.9%. Most of this gain comes from process history rather than contextual attributes, among which Project is the most influential. The study shows what each enhancement technique contributes and where contextual information helps.
Authors: Dahal, Sujan, Alelaimat, Ahmad
Institution: Melbourne Institute of Technology
Language: English
Subfield: Management Information Systems
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7225344126