GCC-Git Change Classifier for Extraction and Classification of Changes in Software Systems

Abstract

Software repositories are used for many purposes like version control, source code management, bug and issue tracking and change log management. GitHub is one of the popular software repositories. GitHub contains commit history of software that lists all changes recorded in the software system, but it does not classify the changes according to the reason for change. In this study a mechanism for extraction and classification of changes is proposed and Git Change Classifier (GCC) tool is developed. The tool uses regular expression to extract changes and employs Text Mining to determine the type of change. GCC Tool reports the year-wise number of changes for a file and classifies the changes into three types: (a) Bug Repairing Changes (BRC), (b) Feature Introducing Changes (FIC) and (c) General Changes (GC). This classification is useful for predicting the effort required for new changes, tracking the resolution of bugs in software and understanding the evolution of the software as it may depend on the type of change.