HETHI - Heterogeneous Hierarchical Knowledge Base

Abstract

The structure of a knowledge base designed by using an extension of the knowledge representation scheme KRP originally developed for the knowledge representation in computer vision systems, is presented in this paper. The HETHI - heterogeneous hierarchical knowledge base consists of one level of the Kanerva-like sparse distributed memory (SDM) and knowledge base designed by the KRP knowledge representation scheme based on the Petri Net theory. The lowest level of the HETH! implemented by Kanerva-like SDM performs associative retrieval information process and supports initialization of the inheritance process at the higher levels of the HETH!. Higher levels of the HETHI are semantic and rule-based levels. In this paper we discuss different reasoning procedures which are supported by the HETHI: "pure" associative infe rence or the so- called recall procedure which is performed only by means of the first level of the HETHI; inheritance procedure defined at semantic level, and finally, inheritance procedure based on the cooperation of the associative recall and inheritance procedure defined at semantic level.