Experimentally Motivated Transformations for Intermodel Links Between Conceptual Models

Complex system development and information integration at the conceptual layer raises the requirement to be able to declare intermodel assertions between entities in models that may, or may not, be represented in the same modelling language. This is compounded by the fact that semantically equivalent notions may have been represented with a different element, such as an attribute or class. We first investigate such occurrences in six ICOM projects and 40 models with 33 schema matchings. While equivalence and subsumption are in the overwhelming majority, this extends mainly to different types of attributes, and therewith requiring non-1:1 mappings. We present a solution that bridges these semantic gaps. To facilitate implementation, the mappings and transformations are declared in ATL. This avails of a common, and logic-based, metamodel to aid verification of the links. This is currently being implemented as proof-of-concept in the ICOM tool.