Multimedia Queries by Example and Relevance Feedback

Abstract

We describe the FALCON system for handling multimedia queries with relevance feedback. FALCON distinguishes itself in its ability to handle even disjunctive queries on metric spaces. Our experiments show that it performs well on both real and synthetic data in terms of precision/recall, speed of convergence, individual query speed, and scalability. Moreover, it can easily take advantage of off-the-shelf spatial- and metric- access methods.

More information

Identifiers

Catalogue record

Export record