Multi-modeling: motivation, concepts and techniques

Slides

Multiple heterogeneous interacting systems are needed to realize the requirements of complex domains. Describing the interactions between these systems and checking their global structural and behavioral consistency is a general, well-known challenge in software engineering. To address this challenge, model-driven software engineering utilizes abstract representations of the constituting systems and their interactions, resulting in a multi-model representing the overall system. In such a multi-modeling setting, global consistency requirements must be satisfied by a set of heterogeneously typed models to guarantee a desired global behavior. If properties of the desired behavior are violated, restoring and propagating restoration of global consistency become challenging issues for developers and model designers. In this talk, I will first give an overall introduction to multi-modelling, then hint at various ways of achieving consistency. If time allows, I will also present an approach to restore consistency based on reinforcement learning.