Research Initiative in Case-Based Reasoning
Abstract
The ability to reason with past cases to solve new problems and justify whether or not a past course of action is, or is not, appropriate and hence should, or should not, be taken is central to many problems of central concern to DARPA, such as: provision of powerful reasoning tools to support high level strategic and tactical planning, understanding complex problem solving behavior, aiding the acquisition of expertise, and improving the actual design and manufacture processes for components of military equipment. For instance, a commander faced with a situation requiring development of a course of action needs to relate this new situation with past situations of a similar nature, analyze what was good or bad about the way the past situations were handled, propose ways perhaps modelled on past solutions to handle the new situation, explore the ramifications and uncover potential fatal weaknesses of these proposals, select the best ones, and explain (particularly to those who must carry it out) and justify (particularly to those in higher command) the chosen course of action. Such decision making often occurs in complex domains, under serious time constraints, and with potentially immense penalties for failure. The best decision makers are both thoroughly knowledgeable about the current situation and ever mindful of the lessons of past history; in a word, they are expert case-based reasoners.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1987
- Accession Number
- ADA249703
Entities
People
- Edwina L. Rissland
Organizations
- University of Massachusetts Amherst