The approach to the representation of imperfect knowledge in expert systems described in this paper is based on the idea that propositions characterizing imperfect knowledge are, for the most part, propositions with implied fuzzy quantities. We consider the propositions as aleatory variables which take their values in the univers E and whose distirbutions can be measured as possibilities of realization. The introduction of the index of fuzziness and the notion of perfect knowledge class the closest of the imperfect knowledge class permit us to represent a common sense of knowledge.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados