Abstract:
One of the most basic phases of the software development life cycle (SDLC) is software
testing. Testing takes up around half of the time needed to release a product. Code-based
testing takes a lot of time and is prone to mistakes. Automated testing by using a Modelbased approach is therefore more favored. The design and development of actual test
cases as well as the priority level classification of the generated test cases are the two
main concerns in software testing. By rapidly locating a small number of test cases that
must satisfy an adequacy condition, automated test case generation techniques attempt to
lower the cost of software testing and contribute to more effective testing of software
products. most of the current test case generation and classification approaches are based
on the system code, even though model-based test case generation methodology which is
the subject of this research work becomes an obvious choice in the software development
industry. Automatic test case Generation and classification remain major issues in
software testing. To fill this gap, we have proposed a model that uses DFS algorithms for
generating the test cases and a machine learning approach used for priority-level
classification of the generated test cases from the UML Activity diagram. The proposed
model for the test case generation tool has been evaluated effectively and it generated all
the possible test cases from activity diagrams by satisfying the test coverage criteria
condition. The study used three machine learning algorithms to classify the generated test
cases based on their priority level and each algorithm achieved an accuracy of SVM
89.73%, KNN 85.63%, and NB 73.93%. as we can see from the results, SVM
outperformed the other two machine learning algorithms.
Key Words: UML Diagram, Test Case, Generation, Classification Prioritization, Activity
Diagram, Activity Dependency Table, Activity dependency Graph, Testing Coverage
criteria.