The quantity of information available on the Web is rocketing, thereby making querying and retrieval of Web data more challenging. These limitations arise because the current Web information systems are lacking particularly in the areas of data relationships and semantics.
Moreover, information on the WWW is placed independently by different organizations; thus, documents containing related information may appear at different Websites and in different formats. Currently, XML is becoming the de facto standard for data publishing and exchange data in most enterprises and institutions. This raises the need for efficient and flexible mechanisms to provide an integrated view over multiple heterogeneous information sources exporting XML documents.
There are two main approaches for providing integrated views over multiple data sources. The first approach called materialized or data warehousing extracts data in advance and stores them in a repository. Then, user queries are addressed directly to the data warehouse system and processed without needing to access the data sources. In the second approach, called mediator systems, views are not materialized, therefore answering queries necessitates the access to the data sources. Both approaches enable users to formulate queries through a mediated schema freeing them from the localization and the heterogeneity of the data.