Transcript Presentation

Future Directions in DOLAP Research
- DOLAP 04 Panel -
Matteo Golfarelli
DEIS – University of Bologna
DOLAP 04
My point of view…
• Ten years of research in DWing and OLAP converged to
define the structure of DW systems (architectures, models,
design, algorithms, etc.)
• These results have been absorbed by vendors to form a
wide set of off-the-shelf software solutions
• Users are now asking for new tools capable of handling
new applications and requirements
Let new applications drive our research!
Considering new applications
• Broaden the idea of DWing …
– Not only multidimensional data….
– Not only extracted in batch mode and analyzed ONLINE by an
expert user
• …without renouncing to its basic concepts
-“…Repository (data & metadata) that contains integrated,
cleansed, and reconciled data from disparate sources for
decision support applications….”
- Information retrieval is not DWing!
• Identify new applications
– Collaborate with researchers and users in different fields (e.g.
economics, life sciences, etc.)
– Keep (present and future) technology advances into account
New applications and challenges I
• Data stream related applications
– OLAP on data streams
– Mining on data streams
– Alerting on data streams
• Challenges
– Architectural issues: adding a real-time integrator and a
main-memory DBMS
– Physical issues: Structures and indexing techniques for real-time
OLAP
– Operational issues: mining algorithms, real-time ETL
New applications and challenges II
• KPI related applications
–
–
–
–
Simulations and WHAT IF analysis on complex KPI graph
Mining KPI patterns
Alerting on KPI
BPM
• Challenges
– Architectural issues: reactive module for handling right-time updates,
Alerting and KPI monitor
– Design issues: methodologies and models for designing data and
processes
– Interface issues: Consider new paradigms for information delivery
– Operational issues: right-time ETL
New applications and challenges III
• DW in life sciences
– Proteins and genome DWs
– Mining on proteins and genome DWs
• Challenges
– Architectural issues: Distributed/federated DWs or P2P DWs
– Interface issues: Complex outputs showing structural protein and genoma
relationships
– Operational issues: ETL on non-relational, strongly heterogeneous DB
New applications and challengesIV
• Complex Data Types applications
– Spatial DWs
– Web data DW
– …
• Challenges
– Design issues: modeling complex data
– Operational issues: tracing data sources on the web
– …
Conclusions
• Classic DW applications and related issues have been
almost completely explored, but…
• … the DWing germ infected the users that are asking for
new applications
• A lot of work is still necessary in the DW field if we accept
to broaden the idea of DWing