An Experimental Investigation into Data Flow Annotated-Activity Diagram-Based Testing
Aman Jaffari, Cheol-Jung Yoo · Journal of Computing Science and Engineering · 2019
With the acceptance of Unified Modeling Language (UML) as the de-facto standard for modeling software systems, many research studies have addressed the necessity for utilizing models of systems under testing as inputs for test automation. Recently, activity diagrams have been used as a basis to derive test cases. Current studies have focused on analyzing the control flow of activities. However, examining the control flow among activities is quite simple and such testing on its own is insufficient. This study proposes technique for test case generation that complements an activity diagram with data flow information. To investigate the potential benefits of this technique, we performed an experimental investigation of well-known systems in testing literature. The experimental results were analyzed and compared with a state-of-the-art test suite generation tool as an alternative approach to fault detection effectiveness and efficiency. Overall, the results indicate that the proposed technique outperforms the alternative approach by detecting 27.3% more faults on average. In particular, the proposed technique yielded the best results in detecting faults related to arithmetic operations or parts used for calculation in our context.