A Computer Program to Learn Production Systems Using a Semantic Net,
Abstract
This paper describes a program that learns production systems. Its productions are designed to produce characterizations of symbol strings. They are learned by having the user present example strings together with the correct characterization. There are two interesting ways to look at the program. On the one hand, the task it is doing is certainly inductive, in the psychologist's meaning of the work. In fact it seems to be a sufficiently powerful sort of induction that it can be used to simulate most other tasks that psychologists would call inductive. Thus the design of the program and the exposition that follows has been influenced by a number of programs that do various specific inductive tasks. On the other hand, since a language for production systems is itself a powerful programming language, this program is doing automatic programming. There are crucial differences between this program and conventional automatic programming systems, but many of the same considerations apply to their design.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1974
- Accession Number
- ADA009142
Entities
People
- Charles L. Hedrick
Organizations
- Carnegie Mellon University