Development of a non-deterministic input-output based relationship test data set minimization strategy

Numerous efficient interaction testing strategies have been proposed in the past literatures to generate optimized test cases for software system under test (SUT). Meanwhile, most of the reported tools enumerate their test cases by covering all t-interactions of parameters involved, which are also known as uniform strength interaction testing. Later on, variable strength interaction testing has been proposed to allow certain subsets to cover higher t-interactions. Nevertheless, this consideration is still insufficient to generate test cases based on actual interactions. In fact, recent researches claimed that neither uniform nor variable strength interaction testing is capable to exactly cover actual factors interaction. Recently, interaction testing based on input-output relationship is reported to accommodate the actual interactions for SUT. This approach claimed that the interaction testing should focus on those input combinations that affect a program output, rather than considering all possible input combinations. The problem of generating minimal test cases is considered NP-complete as no single strategy in input-output interaction testing can generate the most efficient test cases. Therefore, a non-deterministic input-output based relationship test data sets minimization strategy, TIOR, has been proposed in this paper. The benchmarking inputs that reported in the literatures have been executed on TIOR in order to demonstrate its effectiveness and TIOR generally gives competitive results against other strategies.