W3C home > Mailing lists > Public > semantic-web@w3.org > May 2009

Re: Wikitravel.org data

From: Alexandre Passant <apassant@gmail.com>
Date: Sat, 23 May 2009 08:48:34 +0100
Message-ID: <f0e160660905230048w40ac4b42k3c3f5fe23a1ec431@mail.gmail.com>
To: Nicolas Raoul <nicolas.raoul.lists@gmail.com>
Cc: semantic-web@w3.org

On Fri, May 22, 2009 at 9:38 AM, Nicolas Raoul
<nicolas.raoul.lists@gmail.com> wrote:
> Dear all,
> I am thinking of extracting data from wikitravel.org in the same way
> DBpedia extracts data from Wikipedia.
> I am looking for ontologies describing Wikitravel's extractable information:
> 1) Hotels with name, address, phone, prices, services ...
> There seemed to be a travelontology.owl some time ago but now the owl
> file is nowhere to be found.
> 2) Restaurants/bars/clubs with name, address, phone, price range, services ...
> 3) Tourism spots with name, adress, activities ...
> http://rewerse.net/A1/otn/OTN.owl contains some useful concepts but
> more are needed.
> Are there existing ontologies about those domains that I am not aware
> of (and could not find via Google/Swoogle/Sindice) ?

Using the Geonames ontology would probably be the better option [1].

There are various feature code for these purposes, e.g.

  <ontology:Code rdf:about="#S.RSRT">
    <core:definition xml:lang="en">a specialized facility for
vacation, health, or participation sports activities</core:definition>
    <core:inScheme rdf:resource="#S"/>
    <core:prefLabel xml:lang="en">resort</core:prefLabel>

Ideally, you could also provide identity links from your exporter to
Geonames URIs as some features on wikitravel should already be
referenced in Geonames.



[1] http://www.geonames.org/ontology/

> Thanks a lot,
> Nicolas Raoul.
> PS:
> Some parts of http://www.daml.org/ontologies/171 could be used, but I
> find disturbing that all class names and attributes are in German. It
> made me think: If a long-lasting URI string is one that does not let
> people presume what it is about (such as site.org/objects/409824),
> then a long-lasting predicate string URI might be one that follows the
> same principle (such as site.org/predicates/9653). Any thought?
Received on Monday, 25 May 2009 06:17:31 UTC

This archive was generated by hypermail 2.4.0 : Tuesday, 5 July 2022 08:45:12 UTC