Wednesday Feb 28th, 2007
CSE Room 305
12:00 - 1:00 PM
|
Online Estimation For MIN/MAX Aggregate SQL Queries
|
|
Mingxi Wu |
|
Often, the database engine sees similar characteristic queries being asked again and again. It would be very useful to cluster and characterize the historical queries the database engine has seen, and make use of the information at hand to guide the time-consuming query processing. In this talk, we propose a Bayesian framework to process an arbitrary MIN/MAX SQL query in an online fashion. The only assumption under this framework is that the database engine has seen similar characteristic queries in the past. The framework shares all the advantages that the state-of-the-art online database research has: it can constantly maintain a guess to the final answer with statistical meaningful bounds; the guess gets more and more accurate as more and more information is gathered, until it is 100% accurate as the query is completed. In addition, we discuss the broad applicability of this framework to other data management problems.
|
For
upcoming talks, visit http://www.cise.ufl.edu/dbcenter/seminar.shtml