Accession Number:

ADA229122

Title:

Explanation-Based Knowledge Acquisition of Schemas in Practical Electronics: A Machine Learning Approach

Descriptive Note:

Technical rept. 1 Nov 1987-12 Sep 1990,

Corporate Author:

MICHIGAN UNIV ANN ARBOR TECHNICAL COMMUNICATION PROGRAM

Personal Author(s):

Report Date:

1990-09-12

Pagination or Media Count:

79.0

Abstract:

This report describes an AI system that learns electronics concepts from the content of training materials similar to those used in military training in practical electronics. The system is given a series of circuits to learn about each is described with a circuit diagram and a text expressed in propositional form that explains how the circuit accomplishes a specific function. The system uses a naturalistic domain theory to verify the claims made in the text, and then uses explanation-based learning techniques to generalize the explanation and construct a schema for the circuit that can be used to understand later circuits that include the schematic circuits as subcircuits. The system successfully understands later circuits in terms of the schematic ones, and shows savings in some measures of processing as well, and has implications for the design of technical instructional material. Certain important shortcomings of explanation-based learning and schema concepts become clear with this work, and are discussed in some detail. rh

Subject Categories:

  • Personnel Management and Labor Relations
  • Electrical and Electronic Equipment
  • Computer Programming and Software

Distribution Statement:

APPROVED FOR PUBLIC RELEASE