A Comparative Study of Software Bug Clustering Using Lingo and STC Web Clustering Algorithms

Software bug classification is one of the important and popular problems in software engineering. Recently number of algorithms and techniques are presented to automate this process. Software bug data contains number of attributes like bug-id, summary (title), description, comments, status, version etc. Most of the important attributes holds text data. Lingo and STC (Suffix Tree Clustering) both are popular text clustering algorithms used in web mining. In this paper Lingo and STC algorithms are used to classify the software bugs. Classification using clustering methodology is used to create the software bug classes from software bug clusters. In this methodology first clusters are created and then appropriate labels are assigned to the clusters, which indicate the class label for the clusters. Both of these algorithms Lingo and STC are implemented as the part of Carrot2 framework. The software bug repository data is integrated and passed to Carrot2 framework for applying Lingo and STC algorithms. Lingo and STC algorithms are compared for software bug classification task. The comparison is done using various clustering parameters: the number of clusters generated, purity of the clusters and entropy of the clusters created etc.