Automating the creation and handling of software defects

点击翻译按钮获取中文摘要

In software development, similar defects---problems with how the software being built behaves---tend to occur multiple times. Sometimes, it is a defect previously solved for one piece of software that re-occurs in a similar, but new, piece of software. Other times, it is a similar defect that occurs to one previously solved in the same piece of software. Software development projects store records of all defects in defect repositories. These repositories can be mined to learn what information is necessary to solve a defect and to recommend how defects that are similar to the one currently being handled have been solved in the past. The proposed research will extend existing efforts that have learned information from bug repositories by automatically identifying and abstracting concepts from existing defects and by using those concepts to enhance new defect reports and to cluster and find trends in defects that are being prioritized to be solved.