The strength of coalition in a dispersed decision support system with negotiations

Issues related to decision making based on dispersed knowledge are discussed in the paper. A system using a process of combining classifiers into coalitions is used. In this article clusters that are generated using an approach with a negotiation stage are used. Such clusters are more complex and are better able to reconstruct the views of the agents on the classifications. However, a significant improvement is not obtained when we use these clusters without an additional enhancement to the method of conflict analysis. In order to take full advantage of the clustering method, the size and structure of the clusters should be taken into account. Therefore, the main aim of this paper is to examine the impact of the method of conflict analysis and the methods that are used to determine the individual weights of the clusters on the effectiveness of the inference in a system that has a negotiation stage. Four new methods for determining the strength of a coalition are proposed and compared. The tests, which were performed on data from the University of California, Irvine Repository, are presented. The results that were obtained are much better than in the case in which the strength of the clusters was not calculated. The approach that consists in the computation of the individual weights from the judgments of each cluster allowed the size and structure of the clusters to be taken into account. This in turn allowed us to take full advantage of a clustering method with a negotiation stage.