Hybrid Recommendation System Approach for appropriate developer selection in Bug Repositories

Authors : Mohanad ALİMARİ, Sefer KURNAZ, Jalal S H ALBAYATİ

Pages : 471-477

Doi:10.24012/dumf.818164

View : 25 | Download : 12

Publication Date : 2021-06-29

Article Type : Research Paper

Abstract :The essential destination of this research is to develop a hybrid recommendation system methodology to enhance the overall performance accuracy of such existed systems, this recommendation approach normally utilized to assign or propose a few counted numbers of programmers or developers that capable of resolving system's bug reports generated automatically from an open source bug repository, meaning the system decides which programmers or developers should be taken into account to be in charge of finding a solution the bugs mentioned in the bug's report. The definition of the bug selection problems in bug repositories are the activities that developers achieve within program maintenance to fix some specific bugs. Because of lot of bugs are created daily, many developers required are quite large, therefore it is difficult to specify the accurate programmers or developers to find a solution for the issues for specific bug inside the code. The article also aims to improve the accuracy results obtained than existed traditional approaches for this purpose. Besides, we have considered the case of prioritization of system developers, the case can be utilized to find an appropriate grade of developers' achievements as prior knowledge to assist the system in assigning of bug report issue. The results have found that the importance of developers could support the bug triage worker more and help software tasks to solve the bugs fast and within required time during development and support cycles of the software.
Keywords : Open source Bug Repositories, Hybrid machine learning approach, decision trees, Naive Bayes, Random Forest, Neural Networks, feature selection