Talk:Wikidata/Development/RDF

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search

Namespaces[edit]

  • xsd:, rdf:, rdfs:, owl: and prov: with their usual meanings

What is in the rdf: namespace, eg is it RDF markup for a page, expressed in XML? Is there to be a page named rdf:Berlin for instance or maybe one generated from wikidata?

Thanks, --Hypergrove (talk) 20:58, 14 September 2012 (UTC)[reply]

I am sorry that this was not clear. With namespace we mean here the meaning in XML or RDF, i.e. see the RDF spec on namespace. Here this means that the namespaces are simply defined in their usual prefixes, check prefix.cc to resolve them. HTH --denny (talk) 11:03, 19 September 2012 (UTC)[reply]

Semantics of class-class edges[edit]

In https://www.wikidata.org/wiki/Q7868 (cell), there is an edge:

* cell has-part DAPK1-calmodulin complex (source: GO)

But this is misleading. The axiom in GO is

   'DAPK1-calmodulin complex' SubClassOf part-of some cell

This is simplified to a triple in WD. Presumably what happens next is that there is an inference based on an inverse property axiom, giving the edge in the converse direction. But this conflicts with the OWL semantics.

This may or may not matter depending on what the purpose of Wikidata is. If the sole purpose is populating infoboxes with no inferences then this is possibly OK (although if has-part is used to populate an infobox this could result in strange seemingly ad-hoc lists of complexes listed on the 'cell' page). If WD is intended for reasoning then care must be taken here.

Some of the topics here are further expanded in the (incomplete, draft) manuscript here: https://github.com/cmungall/shadowboxing-paper/raw/master/tbox2abox.pdf