Authors:
Jasmin Ramadani
and
Stefan Wagner
Affiliation:
University of Stuttgart, Germany
Keyword(s):
Data Mining, Repository Mining, Coupled Changes, Interestingness.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Knowledge Engineering
;
Knowledge Engineering and Ontology Development
;
Knowledge-Based Systems
;
Service-Oriented Software Engineering and Management
;
Software Engineering
;
Software Process Improvement
;
Software Quality Management
;
Symbolic Systems
Abstract:
Software repositories include information which can be made available for bug fixing or maintenance using
repository mining. The identification of coupled changes have been proposed several times. Yet, existing
studies focus on the found couplings and ignore feedback from developers. We investigate three development
projects and their repositories to find files that frequently change together to support the software developers.
We complement the coupled files information with details from the issue tracking system and the project
documentation. We contrast our findings with feedback from the developers about how interesting our findings
are for them. We found that the small size of the repositories made an insightful analysis difficult. The response
to coupled changes both from experienced and inexperienced developers was mostly neutral. They accepted
most of the additional attributes we presented. Furthermore, developers also suggested other additional issues
to be relevant, e.g. th
e context of the coupled changes and the way they are presented, which we did not
cover in this study. Therefore, coupled change analysis research will need to take the presentation and context
information into account.
(More)