Querying Heterogeneous Information Sources Using Source Descriptions

Alon Y. Levy		Anand Rajaraman		Joann J. Ordille
AT\&T Research		Stanford University	Bell Laboratories
levy@research.att.com	anand@cs.stanford.edu	joann@research.att.com

We witness a rapid increase in the number of structured information
sources that are available online, especially on the WWW.  These
sources store interrelated data on topics such as product information,
stock market information, entertainment, etc. We would like to use the
data stored in these databases to answer complex queries that go
beyond keyword searches.  We describe the Information Manifold, an
implemented system that provides uniform access to a heterogeneous
collection of more than 100 information sources, on the WWW. IM
contains declarative descriptions of the contents and capabilities of
the information sources.  We describe algorithms that use the source
descriptions to prune efficiently the set of information sources for a
given query and practical algorithms to generate executable query
plans.  We also present experimental studies that indicate that the
architecture and algorithms used in the Information Manifold scale up
well to several hundred information sources.