Model-Driven Intelligent Decision Support Systems

In this chapter, the main kind of Model-driven methods that can be used in Intelligent Decision Support Systems are described: Agent-based Simulation Models, Expert-based models, Model-Based Reasoning methods, and Qualitative Reasoning models. Before detailing Agent-based Simulation models, the fundamentals of multi-agent systems are described. Despite there is no unique agent-based simulation methodology in the literature, a general process is described. Next, expert-based models are detailed. As explained in this chapter, expert models are usually coded as a set of inference rules, and hence, the major reasoning technique used is rule-based reasoning. The general architecture of rule-based reasoning is introduced and all main components are analyzed. Special attention is devoted to the Fact Base, Knowledge Base, Reasoning, and Meta-reasoning components. Forward reasoning, backward reasoning, and hybrid reasoning inference engines are detailed, and the corresponding algorithms are presented. Afterwards, model-based reasoning methods are explained. Its foundations and the motivations for their use in IDSS are discussed. As model-based reasoning has been successfully applied in the diagnosis task, the two most used model-based reasoning techniques in diagnosis, consistency-based diagnosis, and abductive diagnosis are explained and algorithms implementing those techniques are presented. Finally, Qualitative Reasoning models are presented. After introducing the fundamentals of qualitative reasoning, and outlining the major tasks in qualitative reasoning (qualitative modelling and qualitative simulation), a general flowchart for deploying these models is detailed. Most commonly used approaches in the literature for both qualitative modelling and qualitative simulation techniques are reviewed. All four kinds model-driven methods are illustrated with practical examples.