Hybrid Knowledge- and Databases

Abstract

In the modern era, databases have been created spanning many domains. However, these databases do not contain general knowledge about their respective domains. For example, whereas a medical database could contain an entry for a patient with some medical disorder, it would not normally contain taxonomic information about medical disorders, known causal agents, symptoms, etc. Collections of this sort of general information are usually called knowledge bases and powerful tools have been developed for querying these collections in complex and flexible ways. The research described in this abstract aims to develop methodologies for merging existing databases with knowledge bases, so that the power and flexibility of knowledge base technology can be applied to existing collections of data.l In traditional DB’s, queries are limited to those that do not reference general information about the domains in which the DB’s are created. To support queries that reference such information, one can merge a DB created in some domain with a knowledge base of the domain to create a hybrid knowledge base / database. Merging a knowledge base with a database extends the range of queries that can be issued. In some cases, ad hoc queries can be simplified when applied to the hybrid KB/DB. Besides expanding the range of queries that can be issued, a hybrid KB/DB also lends itself to research in attribute oriented data mining techniques. In this research, a frame based semantic network, Parka, developed by the PLUS group at the University of Maryland was used to facilitate the merger of a KB with a DB. As an example, we are working with a medical database of about 20,000 OB/GYN patients. The DB is a single table with 70+ columns. Using the original DB, a user could query the DB to list all patients with a particular type of infection. But if the user wanted to query the DB to list all patients with infections known to be caused by any form of bacteria, the query would have to be expressed as a disjunction

Cite

Text

Taylor. "Hybrid Knowledge- and Databases." AAAI Conference on Artificial Intelligence, 1996.

Markdown

[Taylor. "Hybrid Knowledge- and Databases." AAAI Conference on Artificial Intelligence, 1996.](https://mlanthology.org/aaai/1996/taylor1996aaai-hybrid/)

BibTeX

@inproceedings{taylor1996aaai-hybrid,
  title     = {{Hybrid Knowledge- and Databases}},
  author    = {Taylor, Merwyn G.},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {1996},
  pages     = {1411},
  url       = {https://mlanthology.org/aaai/1996/taylor1996aaai-hybrid/}
}