Expert opinions aggregation for discrete events

Abstract

In decision making, a group of experts give opinion on an event say X. There is a need to get a group consensus on X. It is usually not possible to have all experts with the same opinion. Therefore, one needs to fuse different opinions into one opinion (i.e., group opinion). The challenge for this task is the situation that the number of experts is too small, as this situation does not justify the use of the average statistics to come up with a group opinion. This paper addresses this challenge. The main idea of the approach to solve this group decision problem is to consider that the group consensus or opinion is a non-linear function of individual opinions and the non-linear function is further represented by a series of iterations to update the weights in a linear function (i.e., the weighted average of individual opinions). An example is given to illustrate the effectiveness of the approach.