Extending Model-Driven Development Process with Causal Modeling Approach

Abstract

The model-driven development is most promising methodology for cyber-social systems (CSS), cyber-enterprise systems (CES), cyber-physical systems (CPS), and some other types of complex systems. Causality is an important concept in modeling; it helps to reveal the properties of the domain hidden from the outside observer. Great results of CPS engineering based on the perceived causality of specific domain—physical system. The subject domain of the CES as well as of CSS is a complex system type named “an enterprise”. The aim of the article is to enhance a model-based development (MDD) process with a causal modeling approach. The causal modeling aims to reveal the causality inherent to the specific domain type and to represent this deep knowledge on CIM layer. To do this, you need to add a new layer of MDA—a layer of domain knowledge discovery. Traditional MDA/MDD process use the external observation-based domain modeling on CIM layer. Such models assigned to empirical as they based on the notations that do not include causal dependencies, inherent to the domain type. From the causal modeling viewpoint, an enterprise considered to be a self-managed system driven by the internal needs. The specific need creates a particular causal dependence of activities—a management functional dependence (MFD). Concept of the MFD denotes some meaningful collaboration of activities—the causal interactions required by the definite internal need. The first step is conceptualization of the perceived domain causality on CIM layer. A top level conceptual causal model of MFD is defined as a management transaction (MT). The next step is the detailed MT modeling when an elementary control cycle (EMC) is created for each MT. EMC reveals the internal structure of MT and goal-driven interactions between MT internal elements: a workflow of data/knowledge transformations. The results of this study help to better understand that the content of the CIM layer should be aligned with the domain causality as close as reasonable. The main contribution is the extended MDA scheme with a new layer of the domain knowledge discovery and the causal knowledge discovery (CKD) technique tailored for enterprise domain. Technique uses twofold decomposition of management transaction: a control view-based and self-managing view based. The outcome of technique is hierarchy of management transactions and their internal components: lower level management functions and processes, goals, knowledge and information flows. Causal knowledge discovery technique is illustrated using the study programme renewal domain.