Article Preview
TopIntroduction
In the field of software engineering, especially in the phase of software development, there are more chances for the occurrence of errors termed software bugs. Software bugs are also referred to as issues, tickets, and defects. A software bug is an error, flaw, or fault in an application. This error causes the application to produce an unintended or unexpected result, such as crashing or producing invalid results. The primary cause of such bugs is a misunderstanding of requirement elicitation, in other words, if the requirement is not implemented correctly. It has a high impact on the performance of the underlying software project (Hangal & Lam, 2002), which in turn affects the performance of the software organization in its profit and marketing. If the bug emerges during the deployment or testing phase, the first step is to report the bug to the appropriate developer or development team to fix the bugs. However, mostly the development teams are dispersed worldwide. They cannot revisit the error or the bug that arose during the updation of the project. Thus a bug tracking system (BTS) is essential while developing software projects (Birillo et al., 2022; Taha et al., 2021). The BTS keeps track of the reported bugs filed against a software product (Boll et al., 2021; Fouad et al., 2021; Sandhu et al., 2021). It also provides a thorough approach for managing the reported bugs, which contain several information related to a particular bug (Krasner, 2021). The testing team assesses the reported bug in the BTS with different information in the issue report and assigns it to the developer for fixing. This assignment and allocation process of software bugs to appropriate developers is called automatic bug classification (Kim et al., 2006; Anvik et al., 2009; Ahmed et al., 2021; Otoom et al., 2021).
Software companies use various types of BTS to fix reported software bugs efficiently. Companies that develop closed source applications have their BTS designed and implemented according to their needs. On the other hand, companies that develop open-source software projects use bug reports repositories such as Bugzilla (Bugzilla), mantis BT (MantisBT), Hadoop, HBase, HDFS, Mesos, Spark, MapReduce, and Google Chromium (Chromium) for the efficient classification of bugs to respective teams.
The Bug classification process is highly time-consuming, as it must properly allocate the bugs to respective development teams. In addition, manual bug classification is not feasible due to the high number of reported bugs daily. Thus, the manual classification process or reported bugs issues are neither a standard nor a feasible practice for regular practice (Jonsson et al., 2016; Gunawardana et al., 2021; Ramanujam & Padmavathi, 2021) for a large software project. State-of-the-art techniques have proposed various techniques that use bug report description and summary for classification purposes (Shokripour et al., 2013; Tamrawi et al., 2011; Nagwani et al., 2019; Antoniol et al., 2008). However, the bug report contains both the textual and categorical fields that provide various information about a particular bug. Only specific techniques in the existing literature integrate textual and categorical attributes for bug classification. The literature studies are categorized into three main techniques: machine learning (ML), dictionary and information retrieval, and metadata-based approaches. The ML techniques tend to be more effective than the other two approaches (Otoom et al., 2021).