“Knowledge Sifting” for Preliminary Design

Publisher Website
Google Scholar
Abstract
This article describes the use of a new symbolic reasoning methodology, based on the use of matrix associative memories, in the domain of preliminary design. A candidate preliminary design is first formulated in terms of a symbolic constraint system. The symbolic reasoner can then derive information about free variables based on the constraints in the system. One of the major difficulties with current reasoning methodologies in artificial intelligence is the scalability problem; as a problem increases in size, interactions between the elements hinder the reasoning process. Initial investigation indicates that this new symbolic reasoning system provides a means to avoid some of the complications that AI search and logic techniques have when scaling up to real world problems. Also, the methodology appears applicable to both analysis and direct synthesis tasks.

This publication has 0 references indexed in Scilit: