On 6 February at 10:15 Alejandra Duque Torres will defend her thesis "Classifying, Constraining and Ranking Metamorphic Relations“ to obtain the degree of Doctor of Philosophy (in Computer Science).
Supervisors:
Professor Dietmar Pfahl, University of Tartu
Opponents:
Professor Shaukat Ali, Simula Research Laboratory (Norway)
Associate Professor Emelie Engström, Lund University (Sweden)
Summary
Software testing ensures the final software product functions correctly, but it is often time-consuming, costly, and complex, particularly for large systems. One major challenge is the "test oracle problem"—the difficulty of determining the correct output when testing software. This problem arises when the software under test (SUT) lacks an oracle or when creating one is infeasible. Metamorphic Testing (MT) addresses this problem.
Unlike traditional techniques that focus on verifying individual input-output combinations, MT analyzes relationships between input-output pairs across multiple executions. These relationships are defined by Metamorphic Relations (MRs), specifying how outputs should change in response to specific input modifications. Violating an MR signals a potential fault within the SUT, making MT a valuable tool for addressing the test oracle problem. The effectiveness of MT depends largely on the quality of the MRs.
This thesis introduces new methods for classifying, refining, and ranking MRs to improve MT’s efficiency. The first contribution, MetaTrimmer, offers a method for classifying MRs based on their behaviour across different test data inputs. This allows for a more accurate classification by identifying specific cases where MRs are valid. The second contribution, MetaTrimmer+, refines MRs that show mixed results by identifying patterns within the data. This refinement enhances usability and expands the range of effective test cases. The third contribution focuses on ranking MRs based on their defect-detection capabilities, using MetaTrimmer and mutation testing to prioritise the most effective MRs. This method helps reduce the number of test cases while maintaining high defect detection. The contribution three was successfully applied in an industrial case study, proving their practical value.
The defence will be held also in Zoom (meeting ID: 913 8966 7384, passcode: ati).