UOST: UML/OCL Aggressive Slicing Technique for Efficient Verification of Models

摘要

In Model Driven Development (MDD), model errors are a primary concern in development methodology. UML/OCL models have increased both in size and complexity due to its popularity in model design and development. Consequently, the efficiency of the verification process is being affected. The verification of these models is each time more laborious because of their complex design and size thus prolonging the verification process. In this paper, we propose an algorithm of an aggressive slicing technique that works on UML/OCL models (a collection of classes, associations, inheritance hierarchies, and OCL constraints) which improves the efficiency of the verification process. With this technique, the submodels of an original model are computed through partitioning those classes and multiplicities that are not used in written constraints and at the same time, ensuring that the model behavior is not affected. We attempt to quantify the speed-up achieved by adding the slicing technique to two different tools (1) UMLtoCSP and (2) Alloy. The purpose behind showing the results in UMLToCSP and Alloy is to demonstrate that the developed slicing technique is neither tool dependent nor formalism dependent.