In a variety of applications, such as multi-agent systems and heterogeneous knowledge-based systems, there is the need to exchange uncertain information between distinct and independently developed software components. This requires that such components share a common uncertainty interchange format and poses, therefore, a serious and still poorly considered problem, in face of the variety of existing uncertainty theories and of related representation formalisms. In fact, while imposing that all components adopt the same uncertainty theory is often unrealistic and, more importantly, is against the very reasons that support modularization and distribution. Defining a common uncertainty interchange format, able to guarantee compatibility with the different existing approaches (and possibly also with future ones) is an open research problem. In this paper we discuss the basic issues that need to be dealt with for the definition of a sufficiently general uncertainty interchange format and formulate an initial proposal of generic belief representation based on fuzzy numbers.
PDF file