CLUBA: A Clustering-Based Approach for Bug Assignment
Published: 2018
Author(s) Name: Mamdouh Alenezi, Shadi Banitaan and Mohammad Zarour |
Author(s) Affiliation: College of Computer & Information Sciences, Prince Sultan University, Saudi Arabia.
Locked
Subscribed
Available for All
Abstract
Nowadays, software systems are very complex which make software maintenance, especially bug fixing, very challenging. Identifying an appropriate developer to handle a new reported bug is very difficult and error-prone which results in a lengthy bug fixing process. In this paper, we propose a new developer recommendation approach, CLUBA, for assigning relevant developers to fix new bugs. The approach is based on clustering and it recommends a varying list of candidate developers based on their experience. The effect of choosing different percentages of terms from the corpus on clustering quality is carefully evaluated by applying one of the best feature selection methods. Then, similar bug reports are grouped together using the K-means clustering technique. After that, developers are ranked in each cluster based on their experience. We have validated CLUBA on four real open source projects and showed the feasibility of the approach by experimental evaluation.
Keywords: Bug assignment, Developer recommendation, Mining bug repositories, Software maintenance.
View PDF