WrapIt: Automated integration of web databases with extensional overlaps

Abstract

The world wide web does not longer consist of static web pages. Instead, more and more web pages are created dynamically from user request and database content. Conventional search engines do not consider these dynamic pages, as user input cannot be simulated, thus providing often insufficient results. A new approach for online integration of web databases will be presented in this paper. Providing only one sample HTML result page for a source, result pages for new requests will be found by structural recognition. Once structural recognition is established for one source, other web databases of the same universe (e.g. movie databases) can be integrated on the fly by content-based recognition. Thus, the user receives results from various sources. Global schemata will not be produced at all. Instead, the heterogeneity of the single sources will be preserved. The only requirement is given by the existence of an extensional overlap of the databases. © Springer-Verlag Berlin Heidelberg 2003.