Post-consensus analysis of group decision making processes by means of a graph theoretic and an association rules mining approach

It is assumed that a group of experts is tasked to evaluate (rank) a finite set of alternatives during a group decision making (GDM) session. The GDM session may go through a number of iterations (stages) to reach a consensus. At each iteration at least one of the experts changes his/her ranking of some of the alternatives. The session terminates when a consensus has been reached or no expert is willing to alter his/her ranking. In the latter case a compromised consensus is somehow determined. It is also assumed that a data recording mechanism exists that keeps log files with information on each session and the iterations involved at each session. Such logs describe how each individual expert has ranked the alternatives at each iteration of each session. It is also assumed that there is a supervisory authority of this GDM process. This authority wishes to analyze the log files to extract any actionable insights. An approach based on some graph theoretic and the mining of association rules is proposed to identify any dynamics that may exist in the way the experts make ranking decisions. Such analysis may reveal unknown, but potentially useful information, on the way the experts make decisions and also on the way the experts may interact with each other. Knowing such relationships may be pivotal on the way the groups of experts need to be formed and operate during the GDM sessions. Some experimental results based on synthetic data are described and analyzed in terms of the proposed approaches.