Model Engineering

Model Engineering results from the intersection of Model Driven Engineering (mainly from Software Engineering) and Business Modeling (from Industrial Engineering). Building, managing and exploiting models of situations relevant to the CGI research axes.

Scientific objectives

Model Engineering is viewed as the support of an original vision of artificial intelligence within the CGI.

It presents several topics on which the prospective part of the research activities is focused:

Automatic model generation, in particular through the use of learning and data processing technologies.

  1. The automatic or guided evolution of the metamodel (by identification of unused data or by detection of unmapped cognitive space).
  2. The use of situational models with a predictive and anticipatory vocation (by integrating the complexity of the environment and the factors of evolution). 
  3. The use of immersive tools (virtual reality, augmented reality, cockpit and controltower or all types of serious games) to develop both assistance to decision-making and the apprehension of abstract, complex concepts and principles that are not easily visualized ordinarily.

Model engineering is:

  • Modeling situations

This first level corresponds to the construction of accurate, reliable and exploitable models for any situation falling within the scope of the CGI applied research axes (whether a production chain or a complex network of organizations). Such models are systematically obtained by instantiating concepts and relationships represented within situational meta-models, which are intended to describe the domains addressed. The modelling activity itself can be performed manually (by a qualified expert holder of the associated meta-model) or automatically (by different approaches such as the use of learning mechanisms, rule-based systems, or process mining, for example).

  • Managing meta-models

This second level revolves around the definition, construction and updating of meta-models of situations. These meta-models describe business domains (or sub-domains) in the form of sets of concepts, connected by association, inheritance, aggregation or dependency types of relationships. It is then essential to manage the structuring and evolution of these metamodels. The articulation of concepts in terms of genericity or domain specificity, the management of the life cycle and evolution of the meta-model, the principles of identification of emerging concepts, are all topics to which this level is devoted.

  • Exploiting situation models

Schematically, the CGI unit exploits situation models for three distinct and complementary
uses:

  1. the transformation of the knowledge contained within the models to generate additional knowledge (for example in the form of behavior or reaction models adapted to the specificities of the situation encountered);
  2. the visualization of the situation in the form of charts or explicit and exploitable representation elements (cartographies, diagrams, graphs, etc.);
  3. the analysis and research of diagnoses making it possible to understand the specific features of the situation, to anticipate its evolution or to propose a decision-making aid in relation to this situation.