Title: Schemas, Heterogeneity, and Matching.
Speaker: Jayant Madhavan, U. Washington, Seattle.

Abstract:

Schema heterogeneity arises in any application dealing with multiple
sources of data. Large enterprises have to cope with the heterogeneity
arising from the need to import, export, or even continuously
inter-operate between databases having different schemas. Web portals
integrating information from multiple web sites have to cope with the
heterogeneity in the data representations on different web sites. The
need to easily deploy and maintain such systems on a large scale
requires robust automated solutions for coping with schema
heterogeneity.

This talk will focus is on a crucial component of any schema
heterogeneity solution - the task of identifying corresponding
elements in different schemas, otherwise called Schema Matching. I
will describe a schema matching system that is extensible and enables
easy addition and combination of multiple matching techniques. In
particular I will address the following question: can a matching
system acquire expertise from the analysis of a collection of known
schemas and past matching tasks, and use this expertise to better
discover matches between new and unseen schemas? We call this approach
Corpus-based Schema Matching and show that it results in an
improvement in matching, especially for hard to match schema pairs.

I will also describe how such a collective analysis of web services is
used to enhance search in the context of our Woogle web service search
engine.

Bio:
Jayant Madhavan is a PhD candidate at the University of Washington. He
received his Bachelor of Technology degree at the Indian Institute of
Technology, Bombay. His research interests are in the areas of
databases, data mining, machine learning, and information retrieval.
He is particularly interested in information heterogeneity, data
sharing, and the interplay of structured and unstructured data
analysis. His research has developed several schema matching systems,
a search engine for Web Services, and a personal information
management system.

--------------------------------------------------------------------