Researching Of Vertices Rank Model In Dependence Graph Based On Eigenvector Centrality

Modern software design and testing are more complex in the whole software development process with more and more large-scale software engineering. The cost of software testing typically account for more than50%of the total cost of the entire software engineering. So it is an effective solution to improve the efficiency of software testing for reducing software development costs and the time of of software development. With the continuous improving and developing of the software testing techniques, It is very important to use appropriate testing methods based on the actual situation and needs. And it has more practical significant with practical, targeted software testing carrying out. This requires that the program dependence analysis play a important role of software testing process. With the test object through the dependency analysis to discover the characteristics of the application being tested, it makes more specific and targeted by the purpose of software testing. Then it is more efficient to improve the efficiency of software testing with the certain conditions which ensure software procedures perfectly. The current methods of dependence analysis are forward and reverse analysis. And the two methods of dependence analysis are used on the analysis of the entire program or the relationship of a specific point between the other parts of programs. These methods are not effective procedures for the analysis of non-completeblock or unenforceable statement, so this study was to build the vertices rank model of the program dependence graph,which based on the structural characteristics of the program dependence graph. The centrality was used in numbering the relationship of the vertices of software dependence analysis. The Verices Rank Model dependent on the topology of program dependence graph, it is researching of the relationship between the vertices of the dependence graph by quantitative analysis, reflecting the peak in the program dependence graph of the relative importance. The relative relationship of the vertices is more efficient than the dependence of the specific procedures for program dependence analysis. This dependence analysis method using the center of the indicators to measure the relative dependence analysis more objective, subject to the procedures and processes to understand the semantics less affected, while the model for the non-implementation of the statement pieces or block specific programs such as incomplete program dependence graph analysis with good results. The model can be applied to software testing in the unit testing, regression testing, test case generation and test case about the size of reduction. As well as program slicing and chopping, it is usful for the testers in the slice selection criteria and chopping in the identification of variable definition sets and variables reference set.