Improving information retrieval bug localization with query reformulation and haar transform

The daily tasks for Developers are searching source code, to find pieces of code to reuse, to find where to implement changes, in order to fix a coming bugs implied by a specific report, using a query. Locating relevant files depends largely on the quality of the query used. This localization is done with a query formed by a few keywords; the effectiveness of returned results depends strongly on this query, which in case of poor formulation it will lead to biased result. In this paper, we use query reformulation technique to reinforce our results related to bug localization with haar transform (ZOUAIRI & M K ABDI, 2023). Experiments on SWT Eclipse project demonstrate that the algorithms can significantly improve the retrieval performance by 90% averagely, compared to the performance of the original queries given in the benchmarks. Experiments also show that our method can be applied to query expansion and works satisfactorily in selection of good expansion terms.