Authors
V.B. TARASOV
Abstract
The article is devoted to modeling the opinions of a cognitive agent based on both descriptions and prescriptions. Various types of collective opinions are presented. Particular attention is paid to the description of values, ratings and norms in agent theory. In particular, norms are understood as a special case of evaluations. The main place in the work is occupied by a single representation of truth values and modal values within the framework of multi-valued and fuzzy logics. A general approach to modeling various modal systems based on four-valued semantics is proposed. Two types of modality lattices are constructed. In the interests of a formal representation of gradual modalities, formalisms of non-classical measures, in particular, Sugeno measures and Weber measures, are considered. Qualitative measures of possibility and fuzziness based on relations of a non-strict pre-order are investigated. Measures defined on fuzzy sets are also discussed. Relations between degree modalities, fuzzy distributions and non-classical measures are analyzed. This approach, based on both ordinary and fuzzy functions of the set, provides a single representation of the factors of partial truth, uncertainty and inconsistency in the opinions of the agent.
Keywords
cognitive agent, opinion, soft assessment, norm, modality, four-valued semantics, modality lattices, fuzzy set, non-classical measure.
