Login
Computer Science @ Rochester
Friday, April 18, 2003
11:00 AM
CSB 209
Haixun Wang
IBM T.J. Watson Research Center
Weighted Sequences: Pattern Discovery and Language Issues
The database area is experiencing a vigorous growth as a result of a push by new technology combined with a pull by new applications. Advanced applications have placed new demands on database technology for better support. To illustrate these challenges, I will discuss some challenging pattern discovery problems in various application areas, including scientific data analysis, network event management, XML indexing, Bioinformatics, and propose a new technique called weighted sequences to address them.

Then I focus on the problem of incorporating these techniques into DBMSs---a very desirable objective that is difficult to achieve because of SQL's lack of power and extensibility. Thus, I propose minimal extensions to SQL that turn it into a complete language for data-intensive applications, and discuss ATLAS, an experimental DBMS I developed to demonstrate the power and efficiency of these extensions.