Merging Duplicate Bug Reports by Sentence Clustering

Duplicate bug reports are often unfavorable because they tend to take many man hours for being identified as duplicates, marked so and eventually discarded. In this time, no progress occurs on the program in question, and is justifiably an overhead which should be minimized. Considerable research has been carried out to alleviate this problem. Many methods have been proposed for bug report categorization and duplicate bug report detection. However, it is often the case that a duplicate bug report can provide some additional information about a problem which could help in faster resolution of the bug. We propose that duplicate bug reports be merged when possible instead of being discarded, so that maximum information is captured. We propose a clustering-based algorithm to group together similar sentences and create a union of bug reports considered duplicates of each other.