Database Explorer: Large Scale Database Application
Abstract
We describe Forty-Niner (49er), a computer system for automated mining useful knowledge in relational databases, we evaluate 49er on a number of tests, and we describe results of 49er's applications on two NPRDC databases. The basic form of knowledge discovered by 49er is regularities, that is patterns common in sets of data, analogous to scientific laws. In addition, 49er introduces simple concepts. We describe the general form of regularities and we analyze a variety of their types. Regularities discovered in databases can play a role analogous to scientific laws, allowing the users to make prediction, explanations, and justified decisions. There has been considerable interest in recent years in automated methods of mining databases for useful knowledge. Database mining is attractive for many reasons. There are many available databases, they are simple and uniform in structure, and a considerable amount of effort has been spent in designing them and collecting useful data. Many historical records are available in the form of databases, such as stock market prices and indexes, corporate records, census data, and weather records. The automated search for regularities is particularly attractive and useful for large databases, and the same algorithms apply to all relational databases, because of their similarity in structure. Because many databases are very large, and the forthcoming databases will encompass gigabytes or even terabytes of information, knowledge extraction on that scale must be automated.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 24, 1992
- Accession Number
- ADA252183
Entities
People
- Jan M. Zytkow
Organizations
- Wichita State University