New model distances and uncertainty measures for multivalued logic

Abstract

There is an increasing importance of problems regarding the analysis of propositions of experts and clustering of information contained in databases. Propositions of experts can be presented as formulas of n-valued logic Ln. This paper is concerned with defining metrics and degrees of uncertainty on formulas of n-valued logic. After metrics and degrees of uncertainty (as well as their useful properties) have been established, they are used for the cluster analysis of the sets of n-valued formulas. Various clustering algorithms are performed and obtained results are analyzed. Established methods can be further employed for experts propositions analysis, clustering problems and pattern recognition.