W3C home > Mailing lists > Public > semantic-web@w3.org > June 2010

Re: What is it that's wrong with rdf:List [summary]

From: Norman Gray <norman@astro.gla.ac.uk>
Date: Sat, 19 Jun 2010 15:47:47 +0100
Message-Id: <9AEC35B9-26F4-4389-A3A1-03AFD44F1B17@astro.gla.ac.uk>
To: Semantic Web <semantic-web@w3.org>

Greetings.

Many thanks, everyone, for interesting and thoughtful explanations here.

In summary, it seems that:

  * rdf:List isn't fundamentally defective, but
  * it is often at least inconvenient to use with SPARQL, and
  * it collides with OWL, which has hijacked rdf:List for internal use.

SPARQL: the headline practical problem is that there's no way within SPARQL to retrieve all of the (unknown number of) elements of a list, though (as Paul Gearon and Michael Schneider noted) there is talk of SPARQL 1.1 acquiring support for this.  The ontology that I mentioned in the original post <http://www.co-ode.org/ontologies/lists/2008/09/11/> partly addresses this by having a transitive isFollowedBy property (Antoine Zimmerman suggested a similar alternative).  However, as Toby Inkster pointed out, using one of the possible alternative ontologies for this loses out on the relevant syntactic sugar which Turtle provides and which some toolkits support directly (Toby included a useful discussion on some general problems of encoding lists)(.

OWL: it turns out that OWL has rather cheekily hijacked rdf:List for its own purposes, encoding intersections and the like with lists (and that the reasons why this was unavoidable have become slightly obscure).  This means that an ontology which uses rdf:List ends up categorised as being OWL-Full.  It's not clear to me, by the way, under what circumstances things actually break, but it's worth noting (as Pat Hayes does) that a substantial fraction of the RDF and RDFS vocabulary is disallowed in OWL-Lite and OWL-DL <http://www.w3.org/TR/owl-semantics/mapping.html#4.2>.

For my own present purposes, I don't need much in the way of sophisticated reasoning, so don't have to worry too much about OWL niceties in this case.  I suspect I'll end up using rdf:List for its minor syntactic and toolkit conveniences, but with the list elements also containing their own numberings for belt-and-braces goodness.

Best wishes,

Norman


-- 
Norman Gray  :  http://nxg.me.uk
Received on Saturday, 19 June 2010 14:48:18 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 21:45:36 GMT