Transcript Document

Future Directions in
Data Warehousing Research
DOLAP ’04 Panel Discussion
Karen C. Davis
Electrical & Computer Engineering and
Computer Science Dept.
University of Cincinnati
Cincinnati, OH USA
Perspectives Workshop: Data Warehousing
at the Crossroads
Schloss Dagstuhl International Conference and
Research Center for Computer Science
J. Hammer and M. Schneider
(University of Florida)
and T. Sellis
(National Technical University of Athens)
August 1-6, 2004
Seminar 04321
Motivation
• volume of data increases at a staggering rate
• complexity of structure and semantics increases
• representation, manipulation and analysis for novel
applications
Goals
• review state-of-the-art
• discuss recent advances and trends
• identify interesting research problems
Areas for Working Groups
• design and modeling
–
–
–
–
–
–
–
–
–
conceptual modeling
requirements analysis
bridging the gap to data mining
security
metrics
evolution and versioning
interoperability
logical models
design methods
• architecture and processes
Conceptual Modeling
• state of the art: several models proposed for
representing facts, ETL processes, use cases, and
constraints
• challenges: unified, extensible model with formal
semantics
• benefits: CASE tools; wide-applicability of research
results
Quality Metrics
• state of the art: quality models in metadata; normal
forms for DW schemas proposed
• challenges: defining metrics for measuring and
maintaining system quality (both schema and data
quality)
• benefits: better designs and better managed
evolution
Evolution
• state of the art: schema evolution and versioning
proposed in the literature
• challenges: providing effective versioning and data
migration mechanisms to support queries over
multiple versions; propagating changes to ETL
processes
• benefits: avoids data warehouse obsolence and
increases flexibility of queries and what-if analysis
Architecture