A Similarity-based Metamorphic Relations Selection Strategy for Numerical Computation Programs
Song Liu, Shiyu Yan, Xiaohua Yang · 2022
The key to the effectiveness of metamorphic testing in alleviating Oracle problems is the metamorphic relation. For a given numerical calculation program, there are usually multiple metamorphic relations. However, when testing, different orders of metamorphic relations can lead to different overall test effects. How to select a good order of metamorphic relations to achieve the best test results will be an important research direction for metamorphic testing. In this paper, a similarity calculation method between the input relations of the metamorphic relations of the numerical computation program is given, and a similarity-based metamorphic relations selection strategy is proposed. Finally, the proposed strategy is inspected based on the premise of the path coverage adequacy criterion, and the experiment verifies that the strategy is effective and provides a reference for the selection of metamorphic relations.