12 Consensus building in group decision making based on multiplicative consistency with incomplet r

1.题目和关键词
Title:
Consensus building in group decision making based on multiplicative consistency with incomplete reciprocal preference relations
不完全互惠偏好关系下基于乘法一致性的群体决策共识建立
Keywords:
Group decision making (GDM)群体决策;
Incomplete reciprocal preference relations (RPRs)不完全互惠偏好关系;
Multiplicative consistency analysis乘法一致性分析;
Delegation process授权过程;
Feedback mechanism反馈机制.

2.摘要
In this study, a new method is proposed to address group decision making (GDM) using incomplete reciprocal preference relations (RPRs). More specifically, the multiplicative transitivity property of RPRs is first used to estimate missing values and measure the consistency of preferences provided by experts. Following this, experts are assigned weights by combining consistency weights and trust weights. The former are derived by conducting a multiplicative consistency analysis of the opinions of each expert, whereas the latter are used to measure the degree of trust in an expert harbored by others. Experts with satisfactory consistency and large trust weights should typically be assigned large weights. The consensus level is then checked to determine whether the decision making process moves forward to the selection process. If it is negative, a hybrid method consisting of delegation and feedback mechanisms is used to improve the process of arriving at a consensus. The delegation occurs when some experts decide to leave the process, which is common in GDM involving large numbers of participants. The feedback mechanism, one of the main novelties of the proposed approach, generates different advice for experts based on their consistency and trust weights. Finally, a numerical example was studied to show the practicality and efficiency of the proposed method, and the results indicated that it can provide useful insights into the GDM process.

在这项研究中,提出了一种使用不完全互惠偏好关系(RPRs)解决群体决策(GDM)的新方法。更具体地说,首先利用RPRs的乘法传递性来估计缺失值和度量专家提供的偏好的一致性。在此基础上,通过结合一致性权重和信任权重为专家分配权重。前者是通过对每个专家的意见进行乘法一致性分析得出的,而后者则用于衡量其他人对专家的信任程度。具有令人满意的一致性和较大信任权重的专家通常应被分配较大的权重。然后检查一致性程度,以确定决策过程是否前进到选择过程。如果是否定的,则使用由授权和反馈机制组成的混合方法来改进达成一致性的过程。当某些专家决定退出流程时,就会发生授权,这在GDM(群体决策)中很常见,涉及大量参与者。反馈机制是所提出方法的主要新颖性之一,它基于专家的一致性和信任权重为其提供了不同的建议。最后,通过数值算例验证了该方法的实用性和有效性,结果表明该方法可以为GDM过程提供有益的见解。

3.创新点、学术价值
This paper proposes a new hybrid approach for GDM based on consistency analysis under incomplete information. The multiplicative transitivity property of RPRs is first used to estimate missing values and measure the level of consistency. Experts are then assigned different degrees of importance based on consistency weights in conjunction with trust weights. The former is defined to measure the consistency of experts’ opinions, whereas the latter is used to measure the trust relationship between experts. Some delegation and feedback mechanisms to improve the speed of the process are proposed in this study as well. The delegation process is initiated when some experts leave the decision making process, which in turn influences the trust weights of others. The feedback mechanism helps users change their preferences in the direction of greater consensus based on their consistency weights and trust weights.
本文基于不完全信息下的一致性分析,提出了一种新的GDM混合方法。RPRs的乘法传递性属性首先用于估计缺失值并测量一致性级别。然后根据一致性权重和信任权重对专家进行不同程度的重要性分配。前者用来衡量专家意见的一致性,后者用来衡量专家之间的信任关系。这项研究还提出了一些授权和反馈机制来提高该过程的速度。当一些专家离开决策过程时,决策过程就开始了,这反过来又影响了其他人的信任权重。反馈机制帮助用户根据一致性权重和信任权重改变偏好,朝着更一致的方向发展。
(1)该论文使用乘法传递性来估计RPRs的缺失值。与其他基于一致性的方法相比,乘法传递性更适合于RPRs模型的一致性;
(2)基于一致性权重和信任度权重推导出专家的权重,分别用来衡量专家意见的一致性和不同专家的信任度;
(3)授权过程使专家可以离开GDM过程,而不是在这个过程中直到问题结束。此外,由于系统不再需要处理大量的偏好,计算负担也减少了;
(4)反馈机制根据专家的一致性权重和信任度权重为专家生成不同的建议策略。

4.对结论的理解和对学习工作的启发
本文提出了一种不完全RPRs的GDM问题的混合一致性模型。乘法传递性被用来估计缺失值,因为它是建立RPRs一致性模型的最合适的方法。MCI被定义为衡量每个专家的一致性。专家的权重是基于一致性分析和他们对他人的信任度计算得出的。有理由认为,一致性好、信任度高的专家应被赋予较大的权重,以使其相应的意见在聚合过程中能够承载更大的权重。该方法引入了一个授权过程,当有人离开该过程时更新专家信任权重,该方法引入了一个委托过程,当有人离开该过程时更新专家信任权重。此外,如果在任何情况下少数专家都可能是对的,我们将以交互方式提供冲突解决方案,以让所有专家发表意见并与小组意见进行谈判。数值算例表明了该方法的有效性和可行性,并与现有模型进行了比较。可以帮助我们更深入地了解GDM方法的可行性。

Future work:
(1)GDM在整个决策过程中可能涉及到很多因素,如社会心理、政治环境、个人的风险态度等,因此有必要考虑这些因素。
(2)专家在表达他们的偏好关系时可能会表现出某种程度的犹豫。因此,在2类模糊偏好关系和犹豫模糊偏好关系的情况下,开发处理GDM的程序将是有趣的。
(3)共识门槛水平直接影响共识回合,但通常是事先确定的。如何根据不同的因素来确定该参数,例如专家人数、标准或备选方案的数量。
(4)有些专家可能会不诚实地表达自己的观点,或者拒绝改变自己的偏好。因此,可能需要一些新的机制来处理建立共识中的不合作行为。

猜你喜欢

转载自blog.csdn.net/weixin_37996254/article/details/108906932