Identification of Co-change Patterns in Software Evolution
Anushree Agrawal, Rakesh Kumar Singh · 2020
Tracking the after-effect of changes is very difficult due to complex and huge structure of software applications now days. Researchers have found that revision history is very useful in deriving dependencies in software applications. However it is difficult to choose the right history length that is useful for dependency identification. Also all the new and old commits do not have the same potential in identifying dependencies. In this paper, we propose the use of change history for measuring the co-change probability in software applications by varying the weightage of change commits. Academicians and researchers may find this approach useful for studying the effect of revision age on cochange prediction results.