Bug Fix Time Optimization Using K-medoids Clustering Technique

Abstract In software engineering practice, bug triaging, assigning a developer for a bug report to fix it, consumes time besides being error-prone process. Most of previously proposed automation approaches focus on classification techniques. However, the developer workload and developer availability are taken into account in only a handful number papers. In this work, a new automated triaging approach is proposed, which automates the triaging process and optimizes the total fixing time, considering the developer workload, developer availability and the developer previous experience. The proposed solution leverages text mining and clustering technique (K-medoids algorithm). Linux, Eclipse and Apache repositories are explored for evaluating the proposed algorithm. Experimental results showed that the proposed algorithm (1) reduces the total fixing time by 58.16%, 22.84%, and 23.17% in Linux, Eclipse and Apache respectively, (2) normalizes the developer workload and (3) considers the developer availability in assignment decisions.