Software verification is the process that assures that a software product conforms to its predefined specifications. Software inspection is the main verification technique used at early development stages. The inspection effectiveness; number of defects remaining in the artifact relative to number of identified defects; is typically used to measure the quality of the inspection process. Many approaches have been developed to judge the inspection effectiveness such as comparing the inspection results to organization’s historical data, establishing an organizational baseline for defect density, or estimating the number of remaining defects. Towards that end, capture-recapture techniques, originally proposed by biologists for animal population estimation, could be used to estimate the number of remaining defects in a software artifact. This paper reviews the different concepts for defect prediction techniques, provides an elaboration for basic concepts of capture-recapture models with their estimators, its applicability to software defect prediction, state of the art of research for this technique in software inspection and an overall evaluation for capture-recapture technique with its challenges.
展开▼
