A new approach to connective generation in the framework of expert systems using fuzzy logic

A technique for modeling uncertainty in expert systems is presented. Operators are defined using linguistic terms to avoid any numerical representation. These operators consider linguistic term set ordering, constraints that are the counterpart of properties fulfilled by triangular norms in fuzzy logic, and additional restrictions created by the expert´s procedure to combining certainty. The method avoids the usual problems that arise in other treatments of certainty linguistic terms, where they are represented as fuzzy numbers or fuzzy truth labels. One of the most significant problems is the lack of consensus in the representation of each term by a group of experts, due to the necessity of representing the terms in a pseudonumerical scale.<>