A fuzzy-based approach for bug report categorisation. (2017)

• Record Type: Journal Article
• Authors: Chawla, Indu; Singh, Sandeep Kumar
• Is Part Of: International journal of intelligent systems technologies and applications. Volume 16:Number 4(2017)
• Page Start: 319
• Page End: 341
• Publication Date: 2017
• Subjects: ITSs -- issue tracking system -- automated issue labelling -- fuzzy similarity -- open source systems -- software maintenance -- issue reports -- bug repositories; Artificial intelligence -- Periodicals; Intelligent control systems -- Periodicals; 006.3
• Languages: English
• ISSNs: 1740-8865

Abstract

Various studies conducted on bug repositories utilise issue reports labelled as 'bug'. Research conducted on a number of bug repositories have shown that not all issue reports labelled as 'bug' are actually bugs but can also be a request for additional feature, improvement or documentation. This not only threatens the validity of studies that have used mislabelled data but may also give wrong prediction results in future. This has necessitated need for correct labelling of issue reports. The proposed work using fuzzy logic classifier suggests improvement and also reduces the complexity. Validation of this work is done using five open source projects. Experimental results have shown that our approach gives better F-measure scores. The study also elaborated on use of issue reports from other similar projects for training a model; the impact of frequent terms from the training data and applicability of our approach to fine grained categorisation of issues.