File:A dynamic model for political stakeholders - forecasting the actions and relationships of Lebanese Hizbullah with Markov decision processes (IA adynamicmodelfor109455296).pdf

This thesis develops a decision theoretic model rooted in Markov decision process theory to provide military and diplomacy decision makers with insights regarding the interests and potential strategies of Lebanese Hizbullah. State trees are used to capture the interests and actions of Lebanese Hizbullah and other relevant countries, political organizations or group. These state trees are used to design an influence diagram that maps the interdependencies of all interests, actions and players. A Visual Basic for Applications tool was developed for the user to generate the sets of data necessary to populate and solve the model's influence diagram. The actions and interests of Lebanese Hizbullah, over time, in the influence diagram constitute a dynamic Bayesian network. At each stage of this dynamic process, Lebanese Hizbullah is characterized by a state and a set of feasible actions that, depending on the actions taken, determine the transition into a new state of the system. This dynamic dependency-bearing model identifies the most important interests, priorities, and capabilities of Lebanese Hizbullah. The resulting assessment of Lebanese Hizbullah's influence, investment, capabilities, and actions reveal key cause-and-effect relations. The utility of such insights may enable decision makers to determine material variables and best courses of action to enhance their strategic decision-making capabilities.