Constraint-based Model Transformation: Tracing the Preservation of Semantic Properties

Abstract

We present and evaluate a novel constraint based model transformation approach that implements a preservation-centric view. The proposed framework comprises formal preservation constraints that can be used to specify the preservation of invariants that are possibly implemented differently in the source and target model. These invariants are enclosed in concepts, which at the same time serve as grouping mechanism for their different implementations. In that, our framework abstracts from the concrete implementation languages by pre-supposing only a basic set of modeling constructs. To this end, we present two case studies where we apply our approach for the preservation of non-trivial properties and provide some performance analysis where we show that tracking the preservation of a relevant class of complex properties can be done in linear time.