W3C home > Mailing lists > Public > public-wai-ert@w3.org > May 2005

FOAF and Dublin Core terms Re: PRISM standard

From: Charles McCathieNevile <chaals@opera.com>
Date: Mon, 30 May 2005 03:05:50 +0200
To: "Christophe Strobbe" <christophe.strobbe@esat.kuleuven.ac.be>, public-wai-ert@w3.org
Message-ID: <op.srkdb0kdwxe0ny@widsith.local>

Hi Cristophe,

thanks for this - it was a helpful guide to looking at PRISM. Comments  
inline

On Wed, 18 May 2005 18:56:11 +0200, Christophe Strobbe  
<christophe.strobbe@esat.kuleuven.ac.be> wrote:

> PRISM has a 'prism:person' property, which could replace 'foaf:person'  
> (because FOAF is not stable yet). However, 'prism:person' is much  
> simpler than 'foaf:person': it can only contain PCDATA or be empty. If  
> it is empty, you use the 'rdf:resource' attribute. The following  
> examples are taken from the spec [2]:

It seemed to me that this stems from PRISM being conceived as an XML  
specification, not an RDF vocabulary. Effectively I don't think that it  
buys us much - we can also say that foaf:Person either includes PCDATA  
(and then specify actual elements, not just plain text) or use an  
rdf:resource.

The problem with just using plain text is (for example) that people cannot  
spell my name, and it is not obvious from plain text searching that the 6  
or 7 variations on the spelling are all the same person. (I am pretty sure  
that my name is actually unique). Other people write their name  
differently accoridng to circumstances, and automatic completion systems  
tend to produce varieties according to which rules they are following.  
FOAF actually provides us with a way around this.

The benefit of PRISM is that it has formal standardisation status, but I  
think the benefit of FOAF is that it is a better match for EARL, since it  
is designed as RDF already, and the particular parts that seem interesting  
to EARL are the ones that have broad and widely interoperable  
implementation. While the FOAF people are reluctant (since they don't use  
a formal voting process based on some arbitrary membership criteria) to  
nail down their terms forever, the process they have been using for  
several years, along with the evidence of implementation in the real  
world, seems to make it a useful source of terms, if we take some care to  
check each term that we are going to recommend.

Dublin Core is not developed as RDF, but is designed as very broad terms,  
and is designed explicitly to be RDF compatible. And it does have a formal  
process behind its elevation of a term to "recommended", in particular. It  
also has widespread implementation. Because of the DCMI's desire not to  
endorse one syntax over another, it is less interoperable, and in some  
details there are real problems, but again with a little care it seems a  
better source of terms.

In general it seems to me somewhat more risky to take terms from an XML  
vocabulary than from a similar RDF vocabulary, all other things being  
equal. All other things are not, of course, equal. So while in the end I  
recommend that we take terms from FOAF and DC in preference to their  
equivalents in PRISM, without work such as this to check PRISM and other  
specifications it is a little difficult to come to a sensible decision...  
(at least I think that my decsisions make sense to me. Your Mileage May  
Vary, as the americans say :-)

cheers

Chaals

> [1] http://www.prismstandard.org/
> [2] "The PRISM Namespace: Version 1.2", at  
> http://www.prismstandard.org/specifications/1.2/modularized/PRISM_prism_namespace_12.pdf
> [3] http://dublincore.org/documents/dcmi-terms/
> [4] http://www.prismstandard.org/specifications/Prism1%5B1%5D.2.pdf

-- 
Charles McCathieNevile                              chaals@opera.com
          hablo español - je parle français - jeg lærer norsk
   Here's one we prepared earlier:   http://www.opera.com/download
Received on Monday, 30 May 2005 01:06:05 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:18:25 GMT