I am extracting a dataset from Wikivoyage's restaurants, and would like to have some feedback, to know how I can improve it before publication.

For instance, this Wikivoyage restaurant description in Abbeville:

| name=Chez Mel
| alt=Crêperie Mel
| url=http://www.chezmel.fr
| [email protected]
| address=63-65 rue Saint-Vulfran
| lat=50.104437
| long=1.829432
| directions=Near the bridge
| phone=03 22 19 48 64
| tollfree=0800 2133 9843
| fax=03 22 19 48 65
| hours=11:30–14:00, 17:30–23:00
| image=https://commons.wikimedia.org/wiki/File:ChezMel.jpg
| price=10€
| content=Hearty and family-friendly crêpe restaurant.

... becomes this:

<?xml version='1.0'?>
<rdf:Description rdf:nodeID='Nb925d432-69b1-42e0-8063-d914e3504dde'>
    <rdf:type rdf:resource='http://schema.org/Restaurant'/>
    <schema:name>Chez Mel</schema:name>
    <schema:alternateName>Crêperie Mel</schema:alternateName>
    <schema:address>63-65 rue Saint-Vulfran</schema:address>
    <schema:telephone>03 22 19 48 64</schema:telephone>
    <schema:tollfree>0800 2133 9843</schema:tollfree>
    <schema:email>+33 3 22 19 48 64</schema:email>
    <schema:fax>03 22 19 48 65</schema:fax>
    <schema:description>Hearty and family-friendly crêpe restaurant</schema:description>
        <rdf:Description rdf:nodeID='N06aeff0f-6201-482e-bd85-7f4d3c227db5'>
            <rdf:type rdf:resource='http://schema.org/City'/>
            <schema:sameAs rdf:resource='https://www.wikidata.org/wiki/Q28520'/>
<rdf:Description rdf:nodeID="N57a44770-ee4e-4124-9860-661dbf182c1e">
    <rdf:type rdf:resource="http://schema.org/GeoCoordinates"/>

It seems to validate, but how can I improve it?

Also, what other open datasets would it be interesting to reference to?

Context: My goal is to get the data re-used by as many projects/mashups as possible. No such re-users have been identified so far. Another goal is to make it into the LOD graph because it is cool haha.

  • 1
    Shouldn't those GeoCoordinates be somehow linked to the restaurant?
    – svick
    Oct 2, 2014 at 12:21
  • 1
    According to schema.org/Restaurent, 'fax' should be 'faxNumber'. Oct 2, 2014 at 15:06

1 Answer 1


How you can improve it depends very much on how you want to use it, and who you want it to be useful for. 'Valid' RDF is fine, but may not be useful depending on your intended application! If you want it to be helpful to open/linked types, publishing it using other 'accepted' vocabularies is good (although schema.org is the best catch-all, it is not always as expressive as you want):

This also allows any internal applications you have to retrieve extra data from other sources (much as you have done with your wikidata links).

Another thing I'd suggest is that you consider how you're publishing the data, and have a look at the W3C Linked Data Platform principles. The two major points here are:

  1. How you name your RDF resources. Try to give them a URI (in your rdf:resource) that means something rather than just a UUID - this is usually something affiliated with an organisation, or ideally re-using a URI for the same resource if it exists in DBpedia or Freebase.
  2. Publish your data in friendlier formats (not just RDF/XML) unless you have a reason to only concentrate on RDF/XML. It is a horrible format; Turtle is better in every way and easier to work in (but you should provide both, as well as N-triples).

You also need to work out where to publish it to (if you want it out in the open) - this varies depending on what you want to do...

  • Cool. If you explain your use case a bit more, we can probably guide you more effectively as to what approach to take! Oct 7, 2014 at 12:06
  • microformats in html would take care of all of this and be accessible across the web. jon's options are great also, just my two cents
    – albert
    Oct 9, 2014 at 1:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.