Computer Science Seminar

Thursday, October 19, 2006

11-12pm, SB111

Beyond SQL: Structured Data Retrieval by Ranking

Chengkai Li, University of Illinois, Urbana Champaign

Host: Wai Gen Yee

Abstract:

With the massive amount of data everywhere, database systems are facing new challenges: to support non-traditional fuzzy retrieval, in contrast to the Boolean SQL queries, for returning best matches in a ranking of results. That is, even for structured data, we need a retrieval system, much like a ''Google'' for relational databases. Our goal is to support ranking queries, or top-k queries, for matching data by "soft" conditions such as similarity, relevance, or preference, in order to return the best k answers. Such ranking facility can benefit many important database applications, including decision support and OLAP, data warehousing and data mining, document management and text retrieval, E-commerce, multimedia databases, and so on. In this talk, I will introduce our work in building RankSQL, a relational database management system that provides a systematic and principled framework for seamless support of ranking and integration with Boolean query constructs. RankSQL extends relational algebra to treat ranking as a first-class database construct and incorporates the new algebra with the query optimizer and evaluation engine. In addition, we further extend the support of ranking to ad-hoc aggregation queries.

About Chengkai Li:

Chengkai Li is a PhD candidate in the Department of Computer Science, University of Illinois at Urbana-Champaign. His general research interests are in the field of database, with current focus on structured data retrieval and ranking query processing. He also works on Web information integration and XML. Homepage: http://www.ews.uiuc.edu/~cli. Email: cli [at] uiuc [dot] edu.