A tourist is planning a bike trip through northwestern Germany. He plans to cycle along the Weser River and look for some nice locations for fishing and bathing. Information about that river's water quality, therefore, is very important to him. Clearly, this kind of information is somewhere on the Internet, but where?What's the best way to search for geo-referenced data?This scenario is one of the applications under research and development in GeoShare, a cooperative project encompassing the regions of Bremen, Germany; Edinburgh, Scotland; Groningen, the Netherlands; and Sogn og Fjordane, Norway. Our main objective is to develop user-centric services to support better governance, democratic processes, and a sustainable and balanced development of rural and urban areas around the North Sea.Our application for the Semantic Web Challenge 2003 focuses on a tourist looking for information regarding water quality. Here, we'll look behind the scenes at the technologies we used, including ontologies for describing vocabularies and catalogues as well as search mechanisms for keywords with respect to their semantics. The results of the search process can be categorized into two levels of abstraction: professional users of spatially referenced data need to know which server contains the data and in which format, whereas nonprofessionals such as our tourist would prefer digital maps presenting the requested information. The GeoShare project (http:// geoshare.tzi.de) addresses both types of users.
Ontology-based reasoningGeoShare uses the information broker middleware BUSTER (Bremen University Semantic Translator for Enhanced Retrieval), which was developed at the Center for Computing Technologies (TZI) in Bremen. 1 It supports intelligent search and semantic data integration in distributed environments such as the Semantic Web or a geodata infrastructure. It can retrieve distributed information sources based on their conceptual, spatial, and temporal relevance relative to a specific information request. 2 Furthermore, GeoShare lets users integrate heterogeneous information by resolving structural, syntactical, and semantic heterogeneities. In the GeoShare spatial data infrastructure, we use the BUSTER search module as the core component of an Open GIS Consortium-compatible catalog service (see the "Standards and Organizations" sidebar for more details).The search module supports the specification of queries of the type concept @ location in time. In addition to the provision of conceptual semantics-a main requirement of all Semantic Web applications-the system evaluates the spatial and temporal relevance of an information source. 3 For example, a typical information request in our bike tour use case would be "Find up-to-date water-related information regarding the area around Bremen." To reason about conceptual ("concepts equal to or subsumed by the query's concept"), spatial ("regions in or near the requested geographic region"), and temporal relevance ("dates falling within the given period of time"), GeoShare uses metadat...