Abstract: Software defect is important basic data for software quality analysis and improvement.The problems faced by software defect research is how to preprocess noisy defect data effectively before analysis,how to classify defect data accoding to their characters, and how to mine and analyze them. This paper described the contents, methods and technologies of the above mentioned three problems, then compared and analyzed them, at last proposed several problems of defect data which is worth further studying.

Key words: Software defect,Dupficate defect,Defect association,Defect classification,Text mining,Statistical analysis