Building a semantic web based country list

this page contains ideas as a basis for discussion : feedback is needed

Building a authoritive country list

An authoritative list of country names and their alternative renderings in multiple languages can be produced from the Linked Data in GeoNames. Such a list defers maintenance to an external source and can be re-harvested as needs arise.

Loading resources locally

For efficient processing of locality data of mixed quality, we need access to a low latency, fast service that is not rate limited. We propose the maintenance of a local cache of GeoNames RDF that can be dumped and refreshed. For country name lookups, we do not want to load the entire GeoNames RDF store in memory. Instead, we will limit the memory footprint to country names and their alternate renderings in different languages.

Handling misspellings

The RDF cache of country names from GeoNames will not include misspellings. We need to build and maintain our own mapping to enable more robust processing of data of mixed quality. The industry standard for mappings like this appears to be SKOS using a hiddenLabel. Here is an example of a Turtle SKOS file mapping GeoNames country resource to country name misspellings:

@prefix skos: <http://www.w3.org/2004/02/skos/core#> .

<http://sws.geonames.org/6251999/>
  skos:prefLabel "Canada"@en;
  skos:hiddenLabel "canda";
  skos:hiddenLabel "cananda" .

Such a file could be stored in a public version control system allowing the community to validate and contribute to the list.