Qualitative Influence Diagrams With Weights

Along with the rapid development of the social and technology, decision making is becoming more and more important in People's Daily life. As the graph model to describe complex decision-making problems, also a tool for describing and reasoning uncertain decision-making problem, influence diagrams has been widely applied to forecast analysis, decision support, and other areas.When use influence diagrams to represent and decide decision problem, it needs a lot of probability value and accurate utility values. However, it is hard to get these precise, quantitative knowledges. And we only need qualitative knowledge in many practical applications. In order to overcome the shortage of influence diagrams in the practical applications, Wellman proposed the qualitative influence diagrams, Qualitative influence diagrams is the qualitative abstraction of influence diagrams. The influence diagrams and qualitative influence diagrams constructed for the same decision making problems have the same network structure. The different is that qualitative influence diagrams expresses influence relationship of each node using qualitative influence symbols. So, when influence diagram modeling don't need a lot of point probability, only need to get the qualitative relations between nodes, which greatly simplifying the decision-making problems modeling. However, this simple abstract cannot reflect the influence level of the decision-making, also there may appear uncertain result when decision-making. In order to eliminate the uncertainty result produced by decision-making and the influence level of decisions as much as possible, This paper extends qualitative influence diagrams to study the qualitative influence diagrams with weights.First, we introduced the related knowledge and basic concepts of Bayesian networks, the basic concepts of qualitative of Bayesian networks, qualitative relationships and reasoning, the basic concepts and decision-making of influence diagrams and qualitative influence diagrams.Second, we define the weighted qualitative relations of the qualitative influence diagram with weights from probability network and utility network two parts.Third, summarizes the symmetry, transitivity and composition property of qualitative influence diagrams with weights, and shows the proof of addition and multiply operation rules of qualitative influence with weights.Fourth, we extend the decision-making method of qualitative influence diagrams, and show the decision-making method for qualitative influence diagram with weights. The experiments show the correctness and effectiveness of the suggested decision-making method.Finally, it is the summary of our work, meanwhile, we make a prospect for the future work.