W3C home > Mailing lists > Public > public-lod@w3.org > January 2012

Re: Role of RDFa in OWL classes, Systems that create URI's classes but allow regular humans to add information about those classes, alternative classifications

From: Niklas Lindström <lindstream@gmail.com>
Date: Sat, 14 Jan 2012 02:42:22 +0100
Message-ID: <CADjV5jfXJBUZCn_=_9KZdFBw9pSArQp1N=FtL=xsfds=zkm26Q@mail.gmail.com>
To: Peter DeVries <pete.devries@gmail.com>
Cc: public-lod@w3.org
Hello Pete!

On Thu, Jan 12, 2012 at 10:08 PM, Peter DeVries <pete.devries@gmail.com> wrote:
> Hi All,
> I is is my understanding that if we want to setup a web resource that
> defines owl classes you can't use RDFa markup.
> For instance, if I wanted to mint an owl class for the taxon group Mammalia,
> I would need to do that in one of the following ways.
> 1) An OWL ontology created with something like Protege
> 2) A RDF file like http://lod.taxonconcept.org/ses/v6n7p.rdf
> 3) Using one of the various systems like Virtuoso, OpenStructs etc.
> This would be different that defining an instance of the class Taxon for the
> Mammalia in RDFa
> I thought I would just put this to the list to check if my understanding of
> this is correct.

As Gregg said, RDFa can mark up basically any RDF. I wonder though if
something specific has given you the impression that this is not the
case? Admittedly, in RDFa 1.0 there is e.g. no shorthand for
expressing RDF lists (meaning in 1.0 you'd have to use rdf:first,
rdf:next pairs, etc.).  However, (as Gregg also mentioned) RDFa 1.1
has a facility for this (the attribute 'inlist'), along with lots of
other great modifications to improve ease of use and power of
expression. (Among them making RDFa fit better in HTML5 (non-XML),
e.g. by not relying on the use of xmlns.)

Regarding describing classes and such in RDFa, it is so that you're
referring to the various philosophical complications of HttpRange-14,
the meaning of URI fragments in various media types or similar? If so,
I hope that this recent elaboration [1] by the TAG might be of help on
the topic in relation to RDFa.

In part for fun (since I'm intrigued by the data and domain you work
with) and in part to exercise RDFa 1.1 a bit, I created the following
example of your data, to give you an idea of how it can look like
embedded in HTML(5) using RDFa:


Please note that I made this fairly quickly (semi-automatically with
some Vim macros), and that I didn't work on the presentation (at least
some CSS could make it more palatable visually). I did take some care
in expressing all of your data, and I took the liberty of adding some
language and datatype information as I saw fit (albeit guessing on the
Latin). I also tried to use judiciously varied forms of elements and
attribute combinations, to give you a more substantive feel for how
things can be expressed. (Granted, that does make the markup a bit
dense; but it was quite rich data to begin with.) Also, I used some
semantically unstable HTML5 constructs, such as the <time> element
(though I also added an explicit datatype, which might be superfluous
in the future) and some uses of <link> and <meta> in the body.

RDFa 1.1. is now on the verge of Last Call [2]. We very much
appreciate anyone interested to review and give us feedback on it (the
planned timeline is described in [3])!

(Also note that due to the additions and changes in RDFa 1.1 you must
use an RDFa 1.1 aware parser, such as [4].)

Best regards,

[PS. Had I more time I'd elaborate on some ideas of how you might
reduce the number of explicit triples by relying on inference, and
possibly by distinguishing also the information resource describing
the class from its two format representations. Alas, that'd be a much
more involved and open-ended discussion. And I am sure that you are
aware of much of this deeper stuff, as evidenced by the rest of your
mail. So for now I leave this topic by recommending a look at the
(optional) vocabulary expansion part of RDFa 1.1 [5], which I hope is
of interest for anyone using RDF (regardless of format). Also, these
aspects are orthogonal to serialization (though in some respects
editing and consumption are important common denominators).]

[1]: http://lists.w3.org/Archives/Public/public-rdfa-wg/2011Dec/0055.html
[2]: http://lists.w3.org/Archives/Public/public-rdfa-wg/2011Dec/0057.html
[3]: http://lists.w3.org/Archives/Public/public-rdfa-wg/2012Jan/0003.html
[4]: http://www.w3.org/2007/08/pyRdfa/Shadow.html
[5]: http://www.w3.org/TR/rdfa-core/#s_vocab_expansion

> One option would be to write some custom application in Ruby on Rails but I
> thought I would ask the list about other alternatives and what their
> experiences are with them.
> In the future we are looking into creating a system where various clades
> like Mammalia, Felidae etc. are defined and allows less technical editors to
> add and edit attributes that help define those groups.
> One thing to note is that various groups recognize different taxonomic
> clades so the species entailed within one family on DBpedia is often
> different than
> the species entailed in the Uniprot Taxonomy.
> What we would like is to allow different groups to assign a given species
> concept to different classifications depending on their analysis needs.
> We have been able to do something like this with Protege but I think it
> would be better to do this in a more user friendly way.
> We have also been thinking of defining subproperties for skos:broader and
> skos:narrower that could be used to link the taxonomic levels together.
> :broaderTaxon and narrowerTaxon
> This would make it clearer that the link is to a broader or narrower taxon.
> In the past one of the leaders in the LOD community suggested creating a
> separate predicate for these kinds of relations and I am thinking if these
> subproperties would address that suggestion without creating to much
> complication.
> Respectfully,
> - Pete
> P.S. I am currently in Woods Hole MA working the the Eol.org and
> GlobalNames.org. I hope to have a new RDF dump soon that includes links
> between my taxa and the related EoL pages. In the past, this links was
> incomplete. I will update the list when the new RDF dump is available.
> ------------------------------------------------------------------------------------
> Pete DeVries
> Department of Entomology
> University of Wisconsin - Madison
> 445 Russell Laboratories
> 1630 Linden Drive
> Madison, WI 53706
> Email: pdevries@wisc.edu
> TaxonConcept  &  GeoSpecies Knowledge Bases
> A Semantic Web, Linked Open Data  Project
> --------------------------------------------------------------------------------------
Received on Saturday, 14 January 2012 01:43:36 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 24 March 2022 20:29:57 UTC