Publishing deep web geographic data

Abstract

This article introduces a design process, called W-RayS, to describe Deep Web geographic data and to publish the descriptions both on the Web of Data and on the Surface Web. The article also outlines a toolkit that supports the process and discusses an experiment in which the toolkit was used to publish data stored in a large map server. Briefly, to describe geographic data in vector format, the designer should first specify views over the underlying geographic database that capture the basic characteristics of the geographic objects and their topological relationships represented in the vector data. The same idea is applied to raster data, but using a gazetteer or any other geographic database that covers the same area as the raster data. Then, the designer should map the view definitions to an RDF schema, following the Linked Data principles. The descriptions of the geographic data are therefore formalized as sets of RDF triples synthesized from the conventional data. To publish geographic data descriptions on the Web of Data, the designer may decide to materialize the RDF triples and store them in a repository or create a SPARQL endpoint to access the triples on demand. To publish geographic data descriptions on the Surface Web, W-RayS offers the designer tools to transform the RDF triples to natural language sentences, organized as static Web pages with embedded RDFa. The inclusion of RDFa preserves the structure of the data and allows more specific queries, processed by engines that analyze Web pages with RDFa.