Decision-making processes often involve dealing with uncertainty, which is traditionally addressed through probabilistic models. However, in practical scenarios, assessing probabilities reliably can be challenging, compounded by diverse perceptions of probabilistic information among decision makers. To address this variability and accommodate diverse preferences regarding uncertainty, we introduce the Probabilistic Abstract Decision Framework (PADF). PADF offers a structured approach for reasoning across different decision criteria, encompassing the optimistic, pessimistic, and Laplace perspectives, each tailored to distinct perceptions of uncertainty. We illustrate how PADF facilitates the computation of optimal decisions aligned with these criteria by leveraging probabilistic rules. Furthermore, we present strategies for optimizing the computational efficiency of these rules, leveraging appropriate independence assumptions to navigate the extensive search space inherent in PADF. Through these contributions, our framework provides a robust and adaptable tool for effectively navigating the complexities of decision-making under uncertainty.

翻译：决策过程通常涉及处理不确定性，传统上通过概率模型来解决。然而，在实际场景中，可靠地评估概率可能具有挑战性，加之决策者对概率信息的认知存在差异，使得这一过程更加复杂。为应对这种差异并容纳对不确定性的多样化偏好，我们提出了概率抽象决策框架（PADF）。PADF提供了一种结构化的方法，能够针对不同决策准则进行推理，涵盖乐观、悲观和拉普拉斯三种视角，每种视角都针对对不确定性的不同认知进行定制。我们展示了PADF如何通过利用概率规则，促进与这些准则相一致的优化决策计算。此外，我们提出了优化这些规则计算效率的策略，通过利用适当的独立性假设来应对PADF固有的广泛搜索空间。通过这些贡献，我们的框架为有效应对不确定性决策的复杂性提供了稳健且适应性强的工具。