Automatic Attribute-Oriented Fuzzy Induction to Mine Very Large Databases
PI: Dr. Rafal A. Angryk, Department of Computer
Science,
Managing the terabytes of scientific data gathered by
multiple research activities is one of the greatest challenges of the current
decade.� Despite the significant amount
of computational power that all NASA Centers currently have, space-efficient
storage of massive information gathered by spacecrafts during their missions,
as well as time- and bandwidth- efficient transfer, and then processing, of
these resources remain a significant issue.�
We propose to develop a mechanism that will make extraction of knowledge
from large datasets cheaper and more effective than currently available solutions.
In our research we plan to analyze the application of Fuzzy
Concept Hierarchies to reduce large data sets by gradual generalization of
original database records, and to provide data analysts with multi-level views
of the analyzed data.� Our investigation
is based on a popular data mining technique called Attribute-Oriented Induction
(AOI), which utilizes background knowledge,
represented in the form of the trees of abstract descriptors, to generalize the
original information.
We plan to extend the original process of induction with
complete fuzzy generalization models to more accurately model real-life
dependencies among generalized data.� We
want to investigate how the consistent Fuzzy Concept Hierarchies (FCH) can be automatically created from different types of
pre-specified or well-known relations or distance measures for the analyzed
domains.� This is to provide routine
induction mechanisms, which would allow analysts to perform the
Attribute-Oriented Induction without needing to possess expert knowledge.� The goal of this research is to create a
scalable, consistent, and accurate information reporting mechanism that allows
data analysts to transfer the massive datasets, originally stored in a
database, to the resolution (abstraction) needed.
Contact Information
| Mail: | Dr. Rafal A. Angryk |
 |
Department of Computer Science |
| Montana State University | |
| Bozeman, MT 59717 |
| E-mail: | angryk@cs.montana.edu |
| Phone: | (406) 994-4440 |
| FAX: | (406) 994-4376 |
Updated June 19, 2006
msgcwebmaster@physics.montana.edu