期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2015
卷号：78
期号：2
出版社：Journal of Theoretical and Applied
摘要：Automated bug report clustering and classification plays a significant role in managing, assigning, and understanding the bug categories. The most challenging problem in bug report classification is the inadequate amount of labeled dataset. The proposed framework introduces an Ontology-assisted Semi-supervised Clustering Based Classification (OS-CBC) for bug reports amid a small size of the labeled dataset scenario. The proposed approach enriches the data set of the bug report using constructed Bug and Enriched Meta-feature Extraction (BEME) ontology. Semantic constraints based semi-supervised hierarchical clustering (Semantic-HAC) algorithm prioritizes the constraints for clustering the bug reports based on the BEME ontology. The cluster formation of bug reports depends on the transitive dissimilarity and ultrametric distance using ontology-based prioritized constraints. It extends the dataset (stretched) of the bug reports based on the maximum likelihood of the features in the cluster for labeling the unlabeled data. Moreover, the proposed approach categorizes the bug reports of stretched test set under the category of training set label using Multi-label Naive Bayes (MLNB) classifier. The classification technique focuses on the threshold based filtered weight of each term in the training set to improve the accuracy. The proposed OS-CBC approach significantly improves the classification accuracy of the bug reports.
