Mashups over the deep web

Abstract

Combining information from different Web sources often results in a tedious and repetitive process, e.g. even simple information requests might require to iterate over a result list of one Web query and use each single result as input for a subsequent query. One approach for this chained queries are data-centric mashups, which allow to visually model the data flow as a graph, where the nodes represent the data source and the edges the data flow. In this paper we combine the benefits of such an intuitive graphical modeling framework for these chained queries with the large class of Web data sources that are only accessible by filling out forms. These so-called Deep Web sites offer a wealth of structured, high-quality data but pose also several challenges. We identify and address the main challenges and propose an integrated framework for answering chained queries. © 2009 Springer Berlin Heidelberg.