W3C home > Mailing lists > Public > www-rdf-interest@w3.org > May 2004

Re: Bibliographic Record Schema

From: Morten Frederiksen <morten@wasab.dk>
Date: Sat, 29 May 2004 22:01:41 +0200
To: "Richard Lennox" <listserve@richardlennox.net>, <www-rdf-interest@w3.org>
Message-Id: <04052922014101.01131@zonker.mfd-consult.dk>

Hi Richard, all,

On Friday 28 May 2004 23:50, Richard Lennox wrote:
> Over the last acedemic year I have been creating an RDF based data model
> for Bibliographic data.  I now have a version 1.0, provisionally called
> Bibligraphic Record Schema (BRS).  
Looks very good, although of course I have a few comments:

- I like your use of the word "Mantra" and in general the accompanying guide, 
it seems like it should be easy to use.

- The correct namespace URI for marcrel is 
http://dublincore.org/2000/03/13-marcrel# - note the trailing #.

- You might want to present an example before going into the details of 
section 3.2, that would help with the understanding.

- In section 3.2 item 7, foaf:depicts is suggested as an alternative to 
dc:description, but it seems that foaf:depiction should be used instead. 

- In section 3.3 item 1, the FOAF entities like foaf:person should be 
foaf:Person, with the first letter uppercased (and according to the FOAF 
schema it should be foaf:Organization with a z...).

- In section 3.3 item 4, bib:grpName is suggested as being used for the name 
of a group. The property foaf:name could be be used instead. Same goes for 
organisations in item 5.

- In section 3.3 item 5, a bib:orgNumber property is suggested to identify 
the organisation. If this is to be useful, possible as an 
owl:InverseFunctionalProperty, it must be more specific, I assume one can't 
use a Danish CVR number...

- In the examples, the syntax construct rdf:resource must be all lowercased.

- I can see why the pub_details property is needed, but as it is used 
together with e.g. dcterms:isPartOf, it ends up with three property arcs 
between e.g. an article and the journal(s) it appears in. It seems it would 
be a good idea to reduce this "distance" to two, if at all possible.

- Where you use bib:url, I think it would be better to use dcterms:hasFormat, 
and the describe the offered resource in details with dc:format etc. In any 
case, the URI really should not be a literal value - make it a fullblown 
resource to make it possible to make further statements about it.

- I think you may run in to problems with your suggestion for identifiers, 
several of the characters you use have special uses in HTTP URI syntax (which 
they will become subject to if they are to be used as a means of reference).

- You mention "inference" and "inheritance" a few times, to indicate that 
properties can be left out. This should be backed by OWL, as it doesn't hold 
in itself in RDF.

In closing, you might find that your work will be used more, if it is 
assigned a "prettier" namespace URI, perhaps even a W3C one. If you get more 
backers behind this, I imagine it should be possible to convince the chair of 
this group to provide one.

BTW, good to see my work with names being examined and used, I'll get back to 
you with a closer look at that, but I should note that I'm planning on using 
rdf:parseType="Collection", i.e. the new RDF list construct, for the next 
revision.


Regards,
Morten
Received on Saturday, 29 May 2004 15:59:05 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:52:07 GMT