Improvement on ABDOM-Qd and Its Application in Open-Source Community Software Defect Discovery Process

ABDOM-Q_d is a model to describe the characteristics of software defect discovery time-ordered process, such as periodicity, attenuation, oscillation, incremental and discrete. It can help testing participants to evaluate testing quality and to predict testing process with the time-ordered software defect discovery amounts in a good organized software testing process. Due to the poor organization in open-source software community, software defect discovery data show their obvious uncertainties such as mutations and randomness. In the early study of ABDOM/ABDOM-Q_d this kind of process was excluded from the discussion. So it becomes an issue that whether ABDOM-Q_d model can be applied to open-source community software defect discovery process and reveal new natures and characteristics. In order to answer the question, the normalization of b in ABDOM-Q_d were discussed firstly under the curves paradigms and theirs actual significances discussion in conditions b×τ >0 and b×τ <; 0. With the discussion results, a software defect discovery cycle stability coefficient B was proposed and the improved model ABDOM-QB_d with stronger describing capability was established. Then ABDOM-QB_d was applied in a NASA's open-source project, which is a typical representative of open-source projects, to fit the software defect data and a good fitting result was obtained. Finally with the application results the model's applicability and the characteristics of open-source community software defects discovery process were preliminarily discussed.