DATA REDUCTION TECHN IQUES FOR PROFICIENT BUG TRIAGE

IT companies give over 45% of revenue in marketing with software bugs. Bug triage is important step in fixing bugs; its aim is to correctly assign a developer to new bug. To reduce time, text classification methods are adapted to perform automatic bug triage. Here, we concentrate on how to minimize the scale and improvement of quality of bug data. We com bine instance selection with feature selection to reduce data size on the bug dimension and the word dimension simultaneously. Attributes from historical bug datasets is collected for determining the order to be applied for instance and feature selection, predictive model is implemented for a new bug data set. Our contribution results in reduced and high - quality bug data.