Date: November 2, 1995
Speaker: H.V. Jagadish, AT&T Labs, Murray Hill, NJ
Support for Retrieval by Similarity
Abstract:
Databases have traditionally been designed towards retrieval based on the "exact" satisfaction of a query predicate. With the need to store new data types, particularly multimedia data, there is often a need to retrieve items from a database based on some measure of similarity, rather than an exact match. A significant challenge in this context is the design of suitable index structures and retrieval techniques.
I will present an informal discussion of my state of thinking on this topic. I will describe some components of a general framework for solving this problem, and illustrate these with several examples primarily from my recent work on time sequence data, but also some from image and text data.
SDI / LCS Seminar Questions?
Karen Lindenfelser, 86716, or visit www.pdl.cmu.edu/SDI/