The Multilinear Target-based Preference Functions (MTPFs) support multi-attribute decision problems characterized by attribute interactions and targets. However, existing research falls short in flexibly modeling high-dimensional interactions and lacks robustness in decision-making recommendations when faced with uncertain parameters and targets. The paper proposes a robust multilinear target-based decision analysis framework considering high-dimensional interactions, along with uncertainties in parameters and targets. First, the necessity of high-dimensional interactions and the limitations of available MTPFs in modeling high-dimensional interactions are demonstrated. Second, the MTPFs based on the 2-interactive fuzzy measure and the Nonmodularity index are proposed to model the high-dimensional interactions and simultaneously reduce the computational challenges of parameter identification. Third, new descriptive measures are proposed based on the Stochastic Multicriteria Acceptability Analysis to evaluate the robustness of decision recommendations subject to uncertain targets and parameters. The validation and advantages of the framework are illustrated with simulation studies and an application in customer competitive evaluation of smart thermometer patches.

中文翻译：

---

考虑高维交互的稳健多线性基于目标的决策分析


基于目标的多线性偏好函数 （MTPF） 支持以属互和目标为特征的多属性决策问题。然而，现有的研究在灵活地对高维交互进行建模方面存在不足，并且在面对不确定的参数和目标时，决策建议缺乏稳健性。该论文提出了一个稳健的基于目标的多线性决策分析框架，考虑了高维交互以及参数和目标的不确定性。首先，证明了高维交互的必要性和可用 MTPF 在建模高维交互中的局限性。其次，提出了基于 2 交互模糊度量和非模指数的 MTPFs 来模拟高维交互，同时降低参数识别的计算挑战。第三，基于随机多标准可接受性分析提出了新的描述性措施，以评估受不确定目标和参数影响的决策建议的稳健性。该框架的验证和优势通过仿真研究和智能温度计贴片的客户竞争评估中的应用来说明。