An inference model with probabilistic ontologies to support automation in effects-based operations planning

• Ano de defesa: 2012
• Autor(a) principal: Henrique Costa Marques
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Instituto Tecnológico de Aeronáutica
• Palavras-chave em Português: Modelos de decisão; Ontologias (inteligência artificial); Planejamento de processos automatizados por computador; Sistemas de apoio à decisão; Administração; Engenharia de software; Computação
• Link de acesso: http://www.bd.bibl.ita.br/tde_busca/arquivo.php?codArquivo=2190

In modern day operations, planning has been an increasingly complex activity. This is especially true in scenarios where there is an interaction between civilian and military organizations, involving multiple actors in a diversified way, with the intertwining requirements that limit the solution space in non-trivial ways. Under these circumstances, decision support systems are an essential tool that can also become a problem if not properly used. Although this has been widely recognized by the planning and decision support systems communities, there has been little progress in designing a comprehensive methodology for course of action (COA) representation that supports the diverse aspects of the Command and Control planning cycle in Effects-Based Operations (EBO). This work proposes an approach based on probabilistic ontologies capable to support task planning cycle in EBO at the Command and Control tactical planning level. At this level, we need to specify the tasks that will possibly achieve the desired effects defined by the upper echelon, with uncertainty not only in the execution, but also from the environment parameters. Current approaches suggest solutions to the operational level, giving greater importance to the process of targeting while approaches to the tactical level do not take into account the uncertainty present in the environment and actions in their ability to achieve the desired effect. To offer a possible solution to knowledge representation at the tactical level, an inference model was developed to generate the planning problem to be sent to a planning system. The proposed model also describes simulation as a tool to assist the plan';s refinement. The main contribution of this work is the development of a process of probabilistic inference against a knowledge base that is capable of dealing with uncertainty at the tactical level, where different tasks can achieve the same effect, but with different probabilities of success. Obtained results indicate the feasibility of the proposal once valid plans are generated in reasonable time from general orders or requests.