Abstract

We present an approach to selecting data to materialize in Web based information mediators by analyzing multiple factors. An issue in building W eb based information mediators is how to improve the query response time given the high response time for retrieving data from remote W eb sources. We had earlier presented a framework for optimizing the performance of information mediators by selectively materializing data. In this paper we describe our approach for automatically selecting the portion of data that must be materialized by analyzing a combination of several factors, namely the distribution of user queries, the structure of sources and the update cost.

Date

September 22, 1999

Authors

Naveen Ashish, Craig A Knoblock, Cyrus Shahabi