Diaria: an integrated inference engine

Abstract

To date all inference engines have been based on modus ponens which is, of course, a deductive inference rule. However, when making inferences, human beings do not only use modus ponens or any other deductive inference rule, they also employ inductive, abductive and deductive inference rules, as required by the circumstances. In this paper, after having studied the different kinds of inference in AI: deductive, inductive, abductive and retroductive, we present a design of an inference engine which incorporates three of these kinds of inference and uses each of them when most appropriate, depending on the data, information and knowledge with which it is supplied at any time. Thus we have a true data-driven inference system.