The Optimization Of Metamorphic Testing Based On Equivalence Classes Partition

The information contained in successful test cases is fully exploited and applied by metamorphic testing which can effectively solve the Oracle problem in traditional software testing. The two key elements that will impact metamorphosis testing results are the generation of original test case and the selection/generation metamorphic relationship, a test case generation criteria that based on equivalence class partition, ECCEM (Equivalence-Class Coverage for Every Metamorphic Relation), is proposed in this paper, the criteria involves the equivalence classes of program input domain and metamorphic relations, making the selection of the original test case with a certain purpose. To ensure the partition of equivalence classes having a complete coverage, a clipping rule of program execution path called CPC (Class Path Clipping) is proposed, which can detect whether the type of equivalence classes of program input domain have been completely cut. During the metamorphic testing, a new metrics that combines the utilization of test cases and the capability of mutation detection is proposed in order to comprehensively observing the effect of metamorphic testing. ECCEM criteria satisfiability algorithm improves the results of metamorphic testing, however, the number of original test cases cannot be standardized, this paper proposes two iterative metamorphic testing algorithms that are based on ECCEM criteria:ECCEM satisfiability single point iterative algorithm(ECSIST) and ECCEM satisfiability multi-point iterative algorithm (ECMIST), they both effectively reduce the number of generated test cases, thus improving the utilization of test cases and making the generation of test cases without blindness. Finally, based on the principle of composite function, composite metamorphic relations are constructed through a few simple metamorphic relationships in order to ensuring the ability of metamorphic relations while simplifying the generation of composite relationships which will improve the capability of mutation detection, and enhance the effect of metamorphic testing.