Developmental and Operational
Processes for Agent-Oriented Database Navigation for Knowledge Discovery
October 2003
M. Brian Blake, The MITRE Corporation
Andrew B. Williams, University of Iowa
ABSTRACT
Knowledge discovery in databases (KDD) is an area that has become important
to organizations that search for trends and useful information from their raw
database information. KDD can be a tedious and repetitive humandriven process
with respect to extracting the relevant datasets from databases for processing
in the relevant learning algorithms. We investigate an approach where agents can
control the extraction of the data-sets. We show a software developmental process
and paradigm for programming information agents to extract data-sets based on
a methodology we refer to as "extraction hints". We discuss what data
modeling approaches can be used to allow these information agents to be reusable
across various domains and databases. Lastly, using the aviation domain for motivation,
we show the design of an agent architecture toward the further automation of KDD
using agents.
Additional Search Keywords
N/A
|
