KNOBS (Knowledge-Based System) - The Final Report (1982).
Abstract
KNOBS (Knowledge-Based System) was a pioneering demonstration of several capabilities desirable in future military mission planning systems. This system was the first to unite a frame representation for data, rule based inference, English Language understanding using a conceptual-dependency dictionary based parser for question answering and control, and a constraint-based approach to planning. Beyond the first AI system for military planning that worked with a realistically sized data-base, two important contributions of the KNOBS effort were the identification of stereo-typical planning as an important special case of problem-solving which could be automated, and the notion of a mixed initiative where labor was dynamically divided by the user between the user and the machine. This document is a case study of critical decisions that were made during the period 1978 to 1982 in the development of the KNOBS system from the perspective on continuing work in 1986. Some of the decisions led to fundamental new insights (published here for the first time); others were good ideas that just didn't work out as expected. This report is a frank evaluation of both kinds of decisions.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1986
- Accession Number
- ADA174375
Entities
People
- Ethan A. Scarl
- Jonathan K. Millen
- Richard H. Brown
Organizations
- MITRE Corporation