存在检测错误的软件测试方法  

Capture-Recapture Methods With Inspection Errors

作　　者：张丰博[1] 孙海燕[1] 徐昕[1] 

机构地区：[1]北京航空航天大学数学系,北京100083

出　　处：《数学理论与应用》2005年第2期1-10,共10页Mathematical Theory and Applications

摘　　要：软件系统中通常会存有一些错误,而有些错误只有在进一步的开发中才会被检测出来,即使有专门的检测人员也会由于各种原因而不能清除其中全部的错误.前人用Capture-recapture模型来估计软件中含有的错误数量,并且取得了很好的估计值.但是这一模型假设检测行为总是正确的,而本篇文章中将会考虑这种可能性:检测行为错误;并且在本篇文章中被检测的软件将会被分成许多小块,从中取出一块作为研究对象来给出最终的估计值;在文章最后,将会用数据模型来验证文章得出的结论.There are some defects in software systems which can only be d etected in the further development process.Special inspectors can detect some of these defects but often a few remain undiscovered.Several studies have predicte d the number of the software defects using the Capture-recapture model,and have obtained better effects.But this model suppose that inspection is always correct .Method in this paper will consider this possibility:ispection errors.And the wh ole software will be devided into many small parts,then select one partition whi ch can do the conservative estimation to the whole software as the research targ et,in place of detecting the whole software.At the finality,based on the expecti on of a random variable,estimate the number of the defects in the whole software .

关 键 词：软件测试方法 检测行为 软件系统 检测人员 模型假设 数据模型 估计值 文章 e模型 可能性 对象 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]