Dynamic fuzzy knowledge networks: a kind of hybrid model for the individual teaching in ICAI

Abstract

An intelligent computer aided instruction (ICAI) system suitable for individual teaching should at least provide with the following functions: 1) judging student's understanding of the material being taught; 2) preparing different materials and choosing different teaching methods; and 3) correcting students' answers. So an efficient software support for individual teaching activities in ICAI requires the ability to represent inexact knowledge, to carry out approximate reasoning, and to dynamically adjust the rule-base depending on specific situations. We proposed a new model: the dynamic fuzzy knowledge model which is an integration of rule-based, neural network, and fuzzy reasoning techniques. It makes use of the symbolic logic inference at the conceptual level, and neural computing as nonlinear mapping with self-organisation capability at the parameter/data level. The fuzzy concept works as a bridge between two different computing paradigms. The proposed model can be useful not only in the field of ICAI, but also in other applications of intelligent information processing systems.