With feedback and help from the blogosphere we have updated the geonames ontology and fixed some flaws in the corresponding web service. The ontology now includes linked data such as contains (administrative divisions), neighbours (for features with a boundary) or nearby features. The web service is using two URIs to clearly distinguish between Concept (the thing as is) and Document (the document about it).
Concept vs Document
In the Semantic Web an URL is not just an URL. Four uses of an URL can be identified : the name of the URL, the concept the URL stands for (the thing as is), the web location and the document instance. It is important to clearly distinguish between the concept and the document instance about it. In the Geonames Semantic Web we use these URIs for Berlin:
[1] http://sws.geonames.org/2950159/
[2] http://sws.geonames.org/2950159/about.rdf
If you want to express the fact that you are living in Berlin you use the first URI [1], if you want to make a remark about the information geonames has about Berlin then you use the latter [2].
The geonames web server is configured to return a 303 (See Other) redirection response for a request of [1] the URI for the concept and give [2] as the new location of the document. This is one way how the Technical Architecture Group (TAG) of the W3C has decided to resolve the ambiguity between concept and document and this is also how Tim Berners-Lee has asked us to resolve it. The other accepted way to remove the ambiguity is the use of hashes in the URI.
Future Work
Wit the new linked data (contains,neighbours,nearby) the toponyms are now interlinked with each other and the geonames semantic web has become a real Web. With over 6 million toponyms there are over 6 million ways to enter the Geonames Semantic Web. However, there is currently not a single way to leave it once you have entered it. It has become a huge “Black Hole“.
A next step will be to add links to other resources in the geospatial semantic web domain. Bernard is involved in the design and development of a Semantic Web implementation for the French National Institute of Statistics (INSEE) and we will link the Geonames Resources with their corresponding resources in the INSEE Semantic Web. It will thus be possible to follow the link from the Geonames toponym to the INSEE.
What other Geospatial Semantic Web Resources should Geonames link to? We are looking forward to your comments.