A novel framework to prevent manipulation behaviour in consensus reaching process under social network group decision making is proposed, which is based on a theoretically sound optimal feedback model. The manipulation behaviour classification is twofold: (1) ‘individual manipulation’ where each expert manipulates his/her own behaviour to achieve higher importance degree (weight); and (2) ‘group manipulation’ where a group of experts force inconsistent experts to adopt specific recommendation advices obtained via the use of fixed feedback parameter. To counteract ‘individual manipulation’, a behavioural weights assignment method modelling sequential attitude ranging from ‘dictatorship’ to ‘democracy‘ is developed, and then a reasonable policy for group minimum adjustment cost is established to assign appropriate weights to experts. To prevent ‘group manipulation’, an optimal feedback model with objective function the individual adjustments cost and constraints related to the group threshold of consensus is investigated. This approach allows the inconsistent experts to balance group consensus and adjustment cost, which enhances their willingness to adopt the recommendation advices and consequently the group reaching consensus on the decision making problem at hand. A numerical example is presented to illustrate and verify the proposed optimal feedback model.
1 September 2020 in Computers & Industrial Engineering
This article proposes a framework of personalized feedback mechanism to help multiple inconsistent experts to reach consensus in group decision making by allowing to select different feedback parameters according to individual consensus degree. The general harmony degree (GHD) is defined to determine the before/after feedback difference between the original and revised opinions. It is proved that the GHD index is monotonically decreasing with respect to the feedback parameter, which means that higher parameter values will result in higher changes of opinions. An optimization model is built with the GHD as the objective function and the consensus thresholds as constraints, with the solution being personalized feedback advices to the inconsistent experts that keep a balance between consensus (group aim) and independence (individual aim). This approach is, therefore, more reasonable than the unpersonalized feedback mechanisms in which the inconsistent experts are forced to adopt feedback generated with only consensus target without considering the extent of the changes acceptable by individual experts. Furthermore, the following interesting theoretical results are also proved: 1) the personalized feedback mechanism guarantees that the increase of consensus level after feedback advices are implemented; 2) the GHD by the personalized feedback mechanism is higher than that of the unpersonalized one; and 3) the personalized feedback mechanism generalizes the unpersonalized one as it is proved the latter is a particular type of the former. Finally, a numerical example is provided to model the feedback process and to corroborates these results when comparing both feedback mechanism approaches.
26 April 2024 in Engineering Applications of Artificial Intelligence