文档介绍：1000-9825/2005/16(01)0017 ©2005 Journal of Software 软件学报 ,
*
软件关联缺陷的一种检测方法 
景涛+, 江昌海, 胡德斌, 白成刚, 蔡开元
(北京航空航天大学自动控制系,北京 100083)
An Approach for Detecting Correlated Software Defects
JING Tao+, JIANG Chang-Hai, HU De-Bin, BAI Cheng-Gang, CAI Kai-Yuan
(Department of Automatic Control, Beijing University of Aeronautics and Astronautics, Beijing 100083, China)
+ Corresponding author: Phn: +86-10-82315087, Fax: +86-10-82317328, E-mail: yjingtao@,
Received 2004-04-29; Accepted 2004-07-29
Jing T, Jiang CH, Hu DB, Bai CG, Cai KY. An approach for detecting correlated software defects. Journal of
Software, 2005,16(1):17−28. /1000-9825/16/
Abstract: Software engineers always find the state of some defects can influence the detecting rate of other
defects. These defects are correlated defects. This paper gives the definition of correlated software defects, gets
some property of correlated defects by experiment, and then proposes an approach named testing approach with
defects replacement, to detect correlated defects. It also analyzes the capability and efficiency of the approach by
experiment. The data show that the approach is efficient to detect correlated software defects.
Key words: software testing; failure correlation; correlated software defect; random testing; defects replacement
摘要: 软件中的关联缺陷是一种比较普遍的现象,
,通过一个软件实例验证了缺陷的关联关系,提
出了一种缺陷放回的测试方法用来剔除关联缺陷,
据表明,该方法能有效检测软件关联缺陷.
关键词: 软件测试;失效关联;软件关联缺陷;随机测试;缺陷放回
中图法分类号: TP311 文献标识码: A
软件在计算机系统中起着日益重要的作用,而随着计算机系统的广泛应用,软件可靠性问题在软件工程领
域乃至整个计算机工程领域的重要性不断增加, 在这方面已有许多研究. 目前的软件可靠性模型主要
有[1]Halstead 模型、Jelinski-Moranda 模型、Littlewood-Verrall 模型、Cai
都符合下述假设的 1 个或几个[2]:(a) 失效是相互独立的;(b) 发现缺陷立即剔除(及剔除缺陷的时间可忽略不
计);(c) 剔除缺陷过程中不会引入新的缺陷.
*Supported by the National Natural Science Foundation of China under Grant (国家自然科学基金); the Aeronautics
Basic Science Foundation of China und