Consistent Integration of Models Based on Views of Visual Languages

Abstract

The complexity of large system models in software engineering nowadays is mastered by using different views. View-based modeling aims at creating small, partial models, each one of them describing some aspect of the system. Existing formal techniques supporting view-based visual modeling are based on typed attributed graphs, where views are related by typed attributed graph morphisms. Such morphisms up to now require a fixed type graph, as well as a fixed data signature and domain. This is in general not adequate for view-oriented modeling where only parts of the complete type graph and signature are known and necessary when modeling a partial view of the system. The aim of this paper is to extend the framework of typed attributed graph morphisms to generalized typed attributed graph morphisms, short GAG-morphisms, which involve changes of the type graph, data signature, and domain. This allows the modeler to formulate type hierarchies and views of visual languages defined by GAG-morphisms between type graphs, short GATG-morphisms. In this paper we study the interaction and integration of views, and the restriction of views along type hierarchies. In the main result we present suitable conditions for the integration and decomposition of consistent view models. As a running example we use a visual domain-specific modeling language to model coarse-grained IT components and their connectors in decentralized IT infrastructures.