Study Of Class Integration Test Order Based On Class Importance

In the generation of class integration test order,most of the previous studies focused on the number and complexity of test stubs,but did not consider the influence of class importance on class integration test order.The primary purpose of integration testing is to find faults associated with system components.It is important to reduce the cost of test stubs,but the time of detecting inter-class faults and the efficiency of eliminating faults should not be ignored.Important classes have a large fault tendency and are highly destructive when a fault occurs.If the tester ranks important classes in a lower position,the time of fault detection and the cost of fault repair may be affected,and the test efficiency will be greatly affected.In the existing studies,there are few studies on class importance,and the existing research involving class importance only focuses on the impact of class importance on test stub cost,and the impact of class importance on test efficiency is still unknown.To solve this problem,this thesis proposes four kinds of importance measurement methods according to the identification method of critical nodes in complex networks.There are two kinds of local importance: in-degree importance and out-degree importance,and two kinds of global importance: K-core importance and tight importance.Four kinds of importance measurement methods and three kinds of importance measurement methods are added to the class integration test order generation approach based on graph theory.We analyze and compare the performance of seven importance measurement methods on various evaluation indicators.The experimental results show that the connected branch ratios,the potential losses avoided and the test status of important classes may be completely different when the test stub cost is the same.It is necessary to introduce class importance and multi-angle evaluation indicators into class integration test order.In order to solve the problem that the existing studies do not consider the local importance and the global importance comprehensively when measuring the class importance,and the factors considered are relatively simple,this thesis proposes a class integration test order generation approach based on the importance contribution matrix of neighborhood structural hole.In this approach,the importance contribution matrix is constructed by considering the local importance,global importance and structural hole characteristics,and the importance contribution matrix is combined with the feedback adjustment mechanism to generate the class integration test order incrementally.Firstly,class importance is calculated for each class according to the importance contribution matrix.Secondly,the most important class is selected to join the test order.Then,according to the inter-class dependencies that have been added to the test order and the class to be tested,the class importance is dynamically adjusted.Finally,it repeats until all classes are added to the test order.Experimental results show that the cost of test stubs required by this approach is lower than that of some existing approaches.At the same time,this approach can test more important classes first,and the important classes tested first have a greater impact on the programs.In addition,this thesis designs and implements a class integration test order generation tool based on class importance,which generates class integration test orders according to the two approaches proposed in this thesis.This thesis has 35 figures,26 tables,and 95 references.