I am not a developer, nor a librarian. I am one of the volunteers trying to piece together a logical set of relations between the millions of Wikidata items and, in the process probably reinventing the wheel again each day.

On the practical issue of the various 'levels' of works and what Wikipedia is doing about them:

Wikidata is already creating items for Literary works and separate items for Editions of those works (so that our citations can reference a page in a particular edition) and for individual books (because there a hundred or so individual copies of books which are notable enough to have wikipedia articles about them). Like you I am not clear what the difference is between  'expression' and 'manifestation' and which of them corresponds to an 'edition' so we may or may not already be using those concepts. 

We have created specialist properties for linking works to editions which probably sort of match the properties in the FRBR document.

Note that in many cases however the wikidata item about the work also describes the first edition so I guess that doesn't comply with FRBR. That is unlikely to change unless someone comes up with a use case where it causes real problems.

Wikidata is starting to look at data items for free software projects. These are works which are all digital, probably never get printed. They tend to get updated frequently - often having daily updates. It will be interesting to see to what extent the properties and concepts for books are reusable for these works with their very different publication model. We are not trying to do this for wikipedia pages - we only link to the current page and we use a special type of property for that.

'Classes' in wikidata are not coded into the software - they are just items and are only identifiable by the fact that they are the subject of 'subclass of' statements. This means that creating new classes can be done by anyone at any time. Search well first though. It may already exist under a variant label. Search under the 'List of....' label too. Many classes are linked to 'list of' articles in Wikipedia and not all of them have had their Wikidata labels changed yet to reflect their class nature. 

Hope that helps

Joe


On Wed, May 28, 2014 at 3:20 PM, Markus Krötzsch <markus@semantic-mediawiki.org> wrote:
On 28/05/14 15:56, Daniel Kinzler wrote:
Am 28.05.2014 15:05, schrieb Jean-Baptiste Pressac:
Hello,
I am reading the documentation of WikiData where I learned that new properties
could be suggested for discussion. But this means adding knew properties to
WikiData. However, is it possible to use existing RDF vocabularies

Not directly. At the moment, you would just rely on a convention saying that a
given wikibase property is equivalent to a concept from some other vocabulary.

However, we are in the process of allowing claims on properties however. Once
this is possible, you will be able to connect properties to external
identifiers, much in the way data items about people etc are cross-linked with
external identifiers.

This would allow you to model the equivalence between wikidata properties and
other vicabularied. However, the software itself would not be aware of the
equivalence, so it would not be explicit in the RDF representation of data
items. But it would be easy for an external tool that knows how to interpret
such claims on properties to build an appropriate mapping using owl:sameAs or a
similar mechanism.

Daniel is right about this mechanism (but, as I said earlier today, owl:equivalentProperty is the way to go here, not owl:sameAs). However, there is another important point to consider: statements in Wikidata cannot be expressed as single triples in RDF. You need auxiliary nodes for statements to represent qualifiers and references. For details, see our technical report http://korrekt.org/page/Introducing_Wikidata_to_the_Linked_Data_Web

Due to this, you cannot just take external properties and use them to replace Wikidata properties: the RDF version of Wikidata does not have any property that links subjects (items) and objects (values) directly. There are several approaches to get back to single triples (mainly: named graphs and simplified exports); see the technical report for details.

The other issue is that one has to be aware of is that we use properties not just for the main part of a statement, but also for qualifiers and for references. One should be clear about whether an external property applies to all or only to some of these uses. For example, an external property that has "Person" as its domain should never be used in a reference, even if (maybe in an error) somebody has used the Wikidata property in a reference.

We plan to generate the RDF dumps described in the technical report regularly. This would be a possible place for implementing the re-use of external vocabularies. If you are interested in this, you are welcome to join -- basically, one could have a mechanism based on either a hard-coded mapping (in the export code) or based on templates on property talk pages (like constraints now).

Cheers,

Markus



_______________________________________________
Wikidata-l mailing list
Wikidata-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikidata-l