Matching and Abstraction in Knowledge Systems,
Abstract
The first problem is creating what I'll call a 'Knowledge System,' putting into the computer what people have variously called knowledge, or representations of interesting relationships, or expertise, like what a word means, or how it ought to generate inferences. The second one is getting the system to work. The first problem is a human and theoretical limitation; the second is an engineering limitation. And the third problem is a methodological one. Most of the interesting problems that humans solve are not sovled by following a particular algorithm deterministically to some simple solution. Rather, solutions are usually selected from a large set of possible, more or less 'good' answers to a question; that is, a simple question to retrieve some information usually produces a number of partially correct responses, and that produces a requirement to search a set of alternatives for the preferred ones.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1980
- Accession Number
- ADA095425
Entities
People
- Frederick Hayes-roth
Organizations
- RAND Corporation