A Rules-based system for model composition

Model composition is one of the challenges of software systems modeling. As models are usually created independently, they present many types of heterogeneity (syntactic, semantic, local structural and global structural). These heterogeneities cause conflicts when composing models, and may introduce problems and inconsistencies in the resulting model. Indeed, when composing models, to consider that two elements are different when they are equivalent will introduce redundancy. Worst, considering two elements as equivalent when they are different will result in information lost. The purpose of this work is to present an approach of model composition for dealing with conflicts. Our hybrid approach integrates thus all aspects: semantic, syntactic, local structural and global structural in order to detect the right equivalences and right differences. We use ontologies to treat the semantic aspect of the model comparison. We propose a rule-based system called Co-Models. These rules are proposed formally. Finally, our approach is implemented and validated through a use case from banking domain.