A pattern recognition interpretation of implications

Abstract

This paper develops a model of natural human reasoning with linguistic information. The proposed model is based on a new interpretation for a generalized implication clause with linguistic variables [10, 17]. An implication clause along with the modus ponens rule of inference can be used to model a large class of human decision procedures. Here, an implication is interpreted as a rule of classification. Pattern recognition is basically classification of items into some equivalence classes. Thus, it is possible to develop a pattern recognition interpretation for implication clauses. Using the proposed interpretation, a new approach for automating decision procedures is presented. The new model of reasoning is computationally manageable. Moreover, it has the advantage that it enables application of well-known methods used in pattern recognition and clustering techniques to automate the process of human decision-making. One of these applications is a method of reducing the order of an implication clause, i.e., reducing the number of premises in the implication. This is based on feature selection, which is a stage implemented in many pattern recognition systems. The ideas discussed in this paper have been implemented in an expert system for the computer-aided design of men-machine interface. An overview of that system has appeared [13].