Test case prioritization technique using sequence diagram and labeled transition systems in regression testing

Abstract

Model-Based Testing (MBT) utilizes the models of software to generate the test cases. In line with this, Unified Modeling Language (UML) is widely adopted as a modeling support for MBT and UML sequence diagram is one of the most important diagram in the creation of test cases under MBT umbrella (aided by intermediate model). However, MBT method in general tends to generate a large amount of test cases. It is impractical in testing to execute all of the test cases, moreover if the size is large. Also, it has a greater impact on model-based regression testing. Regression testing is a testing process that is applied after software is modified. As a software evolves, some modifications or new features are added to the software. Thus, it also tends to increase the number of test cases. Retesting a large-size of test cases during regression testing stage is even harder since the allocated time and cost are more limited. In order to overcome this issue, a similarity-based selection technique based on Labeled Transition Systems (LTS) intermediate model is introduced. It will select only the subset of test cases that are less similar and has a larger coverage. Nevertheless, this technique still has a drawback. It does not consider the modified parts of the software while selecting the test cases for regression testing. Thus, this technique is against the goal of regression testing, which the test cases supposed to target the modified part of the software. Therefore, a test case prioritization technique is proposed. In the proposed technique, the generated test cases derived from sequence diagram and LTS intermediate model are prioritized for regression testing. Also, the evaluation of both techniques is done based on a set of two case studies. As a result, the proposed technique is able to overcome the issues of original technique by maximizing early coverage of the modified code in regression testing as well as achieve the early fault detection.