Date: April 8, Tuesday 10-12


Title: Schema matching for large-scale data integration

 

Dr. Wensheng Wu, IBM Research center, San Jose, CA

 

Abstract: A data integration system is a system that provides uniform
access to a set of data sources. Such a system largely facilitates the
retrieval of information from multiple sources. An important task in
building data integration system is schema matching, that is, to discover
semantic correspondences or mappings among the elements from different
schemas. While schema matching has been studied for decades, the
scalability issue has not received much attention until recently. In
particular, there are two dimensions of the scale. One is the number of
elements in the schema, and the other is the number of schemas to be
matched. In this talk, I will first present a solution for discovering
topical structures of databases to support the divide-and-conquer
approach to matching large schemas.  I will then describe a
clustering-based matching algorithm for matching a large number of
schemas. I will also briefly describe a new search-driven matching
paradigm for focused integration when a complete integration is not
required. Finally, I will discuss future directions and conclude the
talk.

 

Biodata: Wensheng Wu is currently a postdoctoral fellow at IBM Research
Center. He got his Ph.D. in Computer Science from University of Illinois
at Urbana-Champaign in 2006. His general research interest includes
database, data mining, information retrieval, and Web technology. His
current research focuses on large-scale data integration.