Test Case Update Detection Using Graph Similarity of UML State Machine in Early Development Process

Test case and its data test are part of software testing, which is a crucial phase of software development. Usually, testing is commonly done by explanatory testing and test case-based testing. However, both testing can be very costly and take most of the development time. Moreover, the test case data generated in the early step of development processes could be invalid as its designed algorithm or software might be revised. Therefore, in this study, we proposed a graph similarity of transformed UML State Machine Diagram to justify whether a feature's algorithm data test needs an update or not. We conducted the test on East Java's student admission portal namely PPDB 2017, 2018, and 2019. From the experiment conducted, we found that our proposed method shows promising results with high precision (up to 100%) and 95.6% recall. Thus, our proposed method can be very useful in the early step of development processes because it can save both time and cost since we can detect which features need a new test case data.