Download presentation
Presentation is loading. Please wait.
1
Paper Presentation Prepared by Dindar Öz
‘‘Research Problems in Data Warehousing,, Jennifer Widom Department of Computer Science Stanford University Paper Presentation Prepared by Dindar Öz
2
What is Datawarehousing?
The collection of architecture , algorithms and tools for bringing together selected data from multiple databases or other information sources into single repository which is called “data warehouse”
3
Why do we need Datawarehousing?
4
Our business challenges
Decisions need to be made quickly. Users are businessman not computer experts. Fast increase in database sizes. Increasing importance of business intelligence,and strategy.(Decision Support)
5
What does Datawarehousing offer?
Locating the right information Presentation of information (reports ,graphs) Testing of hypothesis Discovery of information Sharing the analysis
6
Architecture
7
Main Components Wrapper Integrator Information Sources Data Warehouse
8
Main Approaches Lazy(on-demand) Approach Eager(in-advance) Approach
Traditional approach Eager(in-advance) Approach Actually this is datawarehousing.
9
Lazy Approach When Rapidly changing data Requirement of recent data
Unpredictable query requests Large number of information sources
10
Eager Approach When... Query range specified and predictable
Requirement of fast data Information sources are busy(Do not want to be interrupted by DW users too often) Private copies of the clients needed.
11
Research Problems Related with... Wrapper Monitor Integrator
Warehouse Specification Optimization
12
Research Problems/Wrapper
Translation Translating the structure of information sources to that of datawarehouse Change Detection *** Monitoring the Information Sources for changes.
13
Change Detection Strategies
Periodic Full-copy propagation. (Offline) Cooperative Sources (Triggers ,Active Database) Logged Sources (Log Analysis) Queryable Sources (Query Polling) Snapshot Sources (Comparison of Snapshots)
14
Research Problems/Integrator
View Maintenance - Information sources do not care view maintenance.Integrator are loosely coupled with I.S.s - Some of the warehouse views can not be supported by base sources such as historical view of a certain data.
15
Some Optimizations Update Filtering Self-Maintainability
Multiple View Optimization
16
Conclusion Datawarehousing is an indispensible technology and a research area for its numerous benefits. There still open research problems related with Datawarehousing.
17
Any Question?
18
Thank You!
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.