A symbolic model for the representation of uncertain knowledge: theory and applications

The authors show that a formal framework can be proposed within which the expert´s uncertain knowledge can be represented and directly encoded in the knowledge based in a purely qualitative way. Slightly more precisely, a probabilistic-like theory can be elaborated in which, however, the uncertainty degree attached to a given statement is not a real number, but rather a symbolic object in a totally ordered set. A symbolic model for the representation of uncertainty in a knowledge-based system allowing the encoding of the expert´s uncertain knowledge without recourse to any quantification is proposed. The application of this model to a problem in civil engineering leads to the design of an expert system for the statical diagnosis of ancient buildings