Michael Rychener

Production Systems as a Programming Language for Artificial Intelligence Applications Degree Type: Ph.D. in Computer Science
Advisor(s): Allen Newell
Graduated: December 1976

Abstract:

EPAM is a simple model of verbal learning that was developed to simulate certain features of human learning, but it has also turned out to be useful for certain kinds of discriminations in Al programs. This chapter describes a production system for EPAM, featuring the automatic addition of productions by the basic system to represent incremental learning of three-letter nonsense syllables. The design of the network represented by the added productions is discussed and its growth described.

Details of the EPAM production system raise several issues with respect to general EPAM variations and with respect to production system issues such as the right set of production-building primitives. A comparison of the present program to a similar one by Waterman, using a radically different production system architecture, is carried out, highlighting the advantages of the present one.

Thesis Committee:
Allen Newell (Chair)

Joseph Traub, Head, Computer Science Department