Chen Li

1
Data IntegrationThe Good, the Bad, and the Ugly

• Chen Li

2
My story
Doing Data Integration research as a PhD student
3
New adventure as a junior faculty (2001)
Problem data needs to be cleaned!
4
Supporting approximate string matching
Movies starring Schwarrzenger
Find movies with a star similar to
Schwarrzenger.
5
More challenging problems in data cleaning!

• Indexing structures
• Top-k queries
• Approximate joins
• Selectivity estimation of fuzzy predicates

Microsoft SQL Server 2005 supports data cleaning
6
(Good) Lessons

• Talk to our customers
• Find their need and real problems
• Test our solutions in general domains

7
Data Integration Good/Bad/Ugly
8
The Good

• Many many many many many applications!

9
The Bad
10
The Ugly
11
The Future
12
Data ? Data integration
in data integration
13
Demand driven
Customer demand
Data Integration toolkit
14
The next Google for data integration?
Goodigle?

• Not just small toolkit of data integration
• The technique should have enough demand!

15
Information IntegrationThe Good, the Bad, and
the Ugly
Thank you!

• Chen Li

16
One problem Approximate String Searching