Building and Using Large Common Sense Knowledge Bases
Abstract
This project explored the use of qualitative representations, analogical reasoning techniques, and sketching to Create and use large-scale domain theories. We developed and demonstrated qualitative models of economic systems relevant to intelligence analysis, and showed how qualitative spatial reasoning could provide geospatial reasoning relevant for reasoning about military courses of action and trafficability. We started producing a new analogical processing technology, based on our earlier cognitive simulation work on analogical reasoning. We developed techniques that enabled realistic-sized situations to be compared, for automatically extracting and extending cases dynamically based on task constraints from a general purpose knowledge base, and how these techniques could be smoothly integrated with first-principles reasoning. We explored sketching as a means of communication between people and software. We developed new techniques for hosting knowledge bases on top of relational databases.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2001
- Accession Number
- ADA397614
Entities
People
- Ken Forbus
Organizations
- Northwestern University