The goal of the project i3 (eye-cube)
is to extend multidimensional OLAP data
cubes with ad hoc interactive mining primitives.
Multidimensional OLAP products provide an excellent opportunity for integrating mining functionality because of their widespread
acceptance as a decision support tool and their existing heavy reliance on manual, user-driven analysis. Most OLAP products are
rather simplistic and rely heavily on the user's intuition to manually drive the
discovery process. Mining technology can play a fitting role in improving the state of these products.
We propose a suite of extensions in the form of a toolkit attached with a OLAP product that will enable richer, faster answering of queries that are currently
handled through manual exploration.
The following is a list of available operators in the iCube project.
|iDiff: Compact summary of reasons for differences in top-level aggregates|
|iExplore: Watch user browse the data and use that to define information
content of unseen cells.|
|iRelax: Generalize from specific cases to general cases.|
Intelligent Rollups in multidimensional OLAP dataG. Sathe and S. Sarawagi Proc. of
the 27th Int'l Conference on Very Large Databases (VLDB), 2001. |
exploration of OLAP data cubes S. Sarawagi Proc. of
the 26th Int'l Conference on Very Large Databases (VLDB), 2000. |
Intelligent, Interactive, Investigation of OLAP data cubes, S. Sarawagi and
G. Sathe SIGMOD, 2000 (Demo paper).|
idiff: Informative summarization of differences in multidimensional
aggregates., S. Sarawagi.
Data Mining and Knowledge Discovery journal, 4(5):255-276,
differences in multidimensional aggregates S. Sarawagi Proc. of
the 25th Int'l Conference on Very Large Databases (VLDB), 1999. Abstract.|
exploration of OLAP data cubes", S. Sarawagi, R. Agrawal, N. Megiddo.
of the Sixth Int'l Conference on Extending Database Technology (EDBT),
Valencia, Spain, March 1998. PDF
Expanded version available as IBM
Research Report RJ 10102 (91918) , January, 1998. PDF