Aprend: um sistema de aquisição automática de conhecimento a partir de exemplos.

Abstract

The major barrier to building expert systems is the knowledge acquisition problem. This work presents the development of the APREND system. The system is a tool for automatic knowledge acquisition. It uses examples as input data, and is based on inductive inference in order to provide two kinds of production rules: complete rules (with certainty factors in the interval [-1 1], and just one condition). Some criteria for goog rules identification and using are showed. An application on nephrology is reported, in order to illustrate the knowledge acquisition process using APREND.