AVS: An approach to identifying and mitigating duplicate bug reports

In general, software enterprises adopting Error Reporting Management Systems during the production/testing process. The types of information and a large amount of data stored in these systems leads to challenges related to the efficiency of error tracking, such as the presence of duplicate bug reports that hinder productivity. Ideally, a tester should identify a duplicate error report before creating it. In this work, we propose the AVS (Automatic Versatile Search tool), that contributes to the identification of duplicate errors based on Information Retrieval and Text Mining techniques. As proof of concept, we implemented the AVS in the context of the Motorola Test Center (MTC) at the Informatics Center of UFPE. Every search by a new error report candidate is preprocessed. Then, the calculation of similarity between the new report and those available in the database generates a ranked list of similarity. In the end, the results are clustering to produce a more advanced process of identifying duplicate potentials. Experiments carried out on a corpus of about 750,000 reports have revealed the tool's usefulness in identifying duplicate error reports.1