智能随机TBFL方法  

Intelligent stochastic testing-based fault localization approach

作　　者：王蓁蓁[1,2] 

机构地区：[1]金陵科技学院信息技术学院,江苏南京211169 [2]江苏省信息分析工程实验室,江苏南京211169

出　　处：《应用科技》2013年第6期39-45,共7页Applied Science and Technology

基　　金：国家自然科学基金资助项目(61170071);金陵科技学院科研基金资助项目(jit-n-201305)

摘　　要：测试集测试结果对程序错误定位算法(testing-based fault localization,TBFL)没有考虑到程序和测试用例集本身的信息,而智能随机TBFL方法,则是利用随机理论将测试员、程序员关于测试用例和程序的先验知识和测试结果信息结合起来,去定位程序的错误语句.智能TBFL算法主要从人类处理信息的通常过程着眼,把测试员交给开发人员的测试结果(主要是用例是否失败和用例覆盖语句情况)分两个层次处理,首先构造执行矩阵E,利用程序先验分布计算测试集捕获错误的能力等级分布,它相当于人类处理信息首先考察信息的正确性和重要性这一事实;其次构造功效矩阵F,利用前述的测试集能力分布计算程序语句出错可能性的等级分布,它相当于人类处理信息时挖掘信息内涵并对信息加以综合从而做出正确判断这一事实.因此文中提出的算法称之为智能算法,它在理论上是可信的.根据文中所提出的两个有关不同TBFL算法比较标准,并在一些实例上把智能算法与其他一些传统错误定位方法进行对比,发现智能算法的功效比较满意.Because the information on the program and the testing case set themselves are not considered in the TBFL （testing-based fault localization） approach, the author proposes an intelligent stochastic TBFL approach, in which stochastic theory is applied to combine the experiencing knowledge of a programmer on the testing case and program with the testing results, so as to locate program faults. In addition, the approach is on the basis of the usual process by which humanity processes information, the testing results （mainly including if the case fails or not and the sen- tence covered by a case） handed over by a tester to a developer is processed in two tiers. Firstly, an executive matrix E is constructed and a program prior distribution is applied to calculate the level distribution of ability that a testing set captures fault, it is equivalent to such a fact that the correctness and importance of information will he firstly con- sldered when humanity processes information. Secondly, an efficacy matrix F is established; by the utilization of the above-mentioned ability distribution on the testing set, the distribution on the probability of program sentence faults is calculated, which is equivalent to such a fact that the information meaning will be attained and the information will be integrated for correct judgment when humanity processes information. Therefore, the algorithms presented here is called intelligent algorithm, which is believable in theory. Aeeording to the comparison standard of the two different TBFL algorithms mentioned in the paper, and the eomparison between the intelligent algorithm and some other tradi- tional approaches on fault localization, it is found that efficacy of the intelligent algorithm is satisfactory.

关 键 词：错误定位 测试为基础的错误定位 随机错误定位方法 软件测试 智能算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]