Multi-criteria group decision making (MCGDM) deals with decision makers who evaluate alternatives over several criteria. MCGDM problems evolve in tandem with the progress of our society. Such progress has given rise to the large-scale group decision making (LS-GDM) problems in which hundreds of decision makers may participate in the decision process and new challenges to face such as groups’ formation and polarization opinions. Most real world MCGDM problems present changing contexts with uncertainty that cannot be modeled by numerical values. Under these circumstances, the use of linguistic variables and computing with words (CW) processes have provided successfully results. Concretely, the 2-tuple linguistic computational model stands out because its precise linguistic computations and high interpretability. On the other hand, pairwise comparison is a widely used elicitation technique in MCGDM, but a large number of comparisons might lead inconsistent decision makers’ preferences. The Best-Worst method (BWM) reduces the number of pairwise comparisons and the inconsistency in decision makers’ opinions. Several BWM approaches have been proposed to manage linguistic information but none of them take advantage of the 2-tuple linguistic computational process based on the CW approach, which would allow to obtain precise and understandable results. This paper aims to present an extended 2-tuple BWM to reduce the number of pairwise comparisons in MCGDM problems and model the uncertainty associated with them to accomplish accuracy computations and obtaining interpretable results. Moreover, we apply our proposal to LS-GDM scenarios in which polarization opinions and sub-groups identification, ignored from any of BWM proposals, are considered. Finally, the new model is applied to several illustrative MCGDM problems.

多准则小组决策（MCGDM）与决策者打交道，他们根据多个准则评估替代方案。MCGDM问题随着我们社会的进步而不断发展。这种进步引起了大规模的群体决策（LS-GDM）问题，其中数百位决策者可能参与决策过程，并面临着新的挑战，例如群体的形成和两极分化的观点。大多数现实世界中的MCGDM问题都呈现出不确定的变化环境，而这些不确定性无法通过数值来建模。在这种情况下，语言变量的使用和带字的计算（CW）过程已成功提供了结果。具体地，二元语言计算模型由于其精确的语言计算和高解释性而脱颖而出。另一方面，成对比较是MCGDM中广泛使用的启发技术，但是大量比较可能会导致决策者的偏好不一致。最佳-最差方法（BWM）减少了成对比较的次数以及决策者意见的不一致。已经提出了几种BWM方法来管理语言信息，但是它们都不利用基于CW方法的二元组语言计算过程，这将允许获得精确且可理解的结果。本文旨在提出一种扩展的2元组BWM，以减少MCGDM问题中成对比较的次数，并对与它们相关的不确定性进行建模，以完成精度计算并获得可解释的结果。而且，我们将我们的提案应用于LS-GDM方案，其中考虑了BWM提案中忽略的极化意见和子组标识。最后，将新模型应用于几个说明性的MCGDM问题。