Software Test Data Generation And Defect Detection Based On Dominant Relationship Between Statements

The final goal in software testing is to find the existing defects or efforts,improving the quality of the software. One of the most difficult aspects in softwaretesting is the generation of effective test data through targeted theory and method.There are large numbers of research achievements on the generating test data anddetecting defects. However, for the special software, such as the software with flagvariables, the efficiency of using traditional method is very low. The study hasshowed that each statement exists closely restriction relationships in the software, andthe dominant relationship between statements is one of the common restrictionrelationships. This paper applies the dominant relationship between statements tosolve the test data generation and defect detection problem for software with flagvariables, and develops the relevant software testing system.For the problem of generating test data of software with flag variables, a test datageneration method based on the dominant relationship between statements ispresented. First, in view that a target statement has several dominant statements, fourindicators for evaluating the coverage difficulty of a statement are presented. Then,the approaches to calculating the values of the above indicators are given, and theeasiest covered dominant statement is chosen using the Topsis sorting. Finally, amathematical model for generating test data problem based on the dominantrelationship between statements is established, and the model is solved by geneticalgorithm, so as to generate the desired test data. The proposed method is applied totest several benchmarks and industrial programs, and the experimental results showthat the proposed method can greatly improve the efficiency of generating test data.For the problem of detecting defects of software with flag variables, a new defectdetection method is presented. First, the principle of detecting defects based on thedominant relationship is provided, and then a mathematical model for generating testdata with the ability of detecting defects is established. Finally, a genetic algorithm isemployed to solve the above model so as to generate the desired test data. Theproposed method is applied to test several benchmark and industrial programs, andcompared with an existing defect detection method. The experimental results showthat the proposed method achieves a higher ability of detecting defects and avoids thetest Oracle problem. Using the above theories and methods, a software testing prototype system basedon the dominant relationship between statements is developed to realize two basicfunctions of generating test data and detecting defects based on the dominantrelationship between statements. The system establishes man-machine interface of testdata generation for detecting defects and handles a user’s input information accordingto her/his test requirements. In addition, the system also provides a user with severalalternate functions, especially about genetic operations. Benchmark programs are runon the prototype system, and the experimental results show that the developed systemis stable and effective, providing technical support for automated software testing.This study based on the dominant relationship between statements generates testdata and detects defects, providing new ideas and improving the effect and efficiencyin software testing. Therefore, it is of important application values.