Classification tree method with parameter shielding

Abstract

The Classification Tree Method (CTM) is a structured and diagrammatic modeling technique for combinatorial testing. CTM can express the notion of “parameter shielding”, the phenomenon that some system parameters become invalidated depending on another system parameter. The current form of CTM, however, is limited in its expressiveness: it can only express parameter shielding that depends on a single parameter. In this paper, we extend CTM with parameter shielding that depends on multiple parameters, proposing CTM. We evaluate the proposed extension on several industrial systems. The evaluation finds that parameter shielding often depends on multiple parameters in real systems, and the effectiveness of the extension.