The IPTC's metadata requirements

Apologies for not responding to the various mails. I hope to get 
around to them.

The IPTC News Metadata Framework Requirements spec (draft), dated 
20 April 2005, is available at:
http://www.iptc.org/download/public/DRAFT-NAR_1.0-spec-NMDF-BusinessRequ
_26.pdf

Apologies for the long URI and for the likelihood of it wrapping.

Apologies also for the almost complete absence of examples.  Pending 
the availability of a more complete document, here is a fantasy 
syntax we've been exploring.  I hasten to add that we do *not* 
propose that this be directly supported by XHTML.  We do hope that 
GRDDL will be able to transform all of this into suitable triples.
Here goes:

   <foo
      val="scheme:code"          <!-- Value -->
      type="scheme:code"         <!-- Concept type -->
      qual="scheme:code"         <!-- Qualifies -->
      parent="scheme:code"       <!-- A parent -->
      why="scheme:code"          <!-- Why present -->
      equiv="scheme:code"        <!-- Equivalent to -->
      assign="scheme:code"       <!-- Assignee -->
      conf="50"                  <!-- Confidence (%) -->
      rel="50"                   <!-- Relevance (%) -->
      when="2005-05-11T12:34:56" <!-- Date and time -->
      xml:lang="en"              <!-- Refers to the Description -->
   >Hello world</foo>            <!-- Description -->

where:

   val identifies the concept, eg Angola or Interview or Japanese or 
       Pharmaceuticals or Tony Blair.  Note that the concept is *not*
       limited to being a *subject*.  Subjects are simply the most 
       obvious example.

   type identifies the concept type, eg Country or Genre or Language 
       or Business sector or Person.

   qual identifies another [umm*] being qualified by this concept, 
       eg "Men's" qualifying "Swimming".

   parent identifies a concept that is the parent of this concept.

   why identifies the reason why the code was addded, eg:

       -  Direct:   A concept which is directly extracted from the 
                    content by a tool and/or by a person (eg Paris 
                    or GlaxoSmithKline)

       -  Ancestor: An ancestor of some other concept (eg the 
                    concepts France and Europe are ancestors of the 
                    concept Paris)

       -  Derived:  A concept derived by look-up in some taxonomy 
                    (eg the concept Pharmaceutical Industry Sector 
                    may be derived from the concept GlaxoSmithKline)

   equiv identifies a semantically equivalent code.

   assign identifies the person/system which will assign or has 
          assigned the code.

   conf holds the Confidence level.

   rel holds the Relevance level.

   when holds the timestamp of when the code was added.

   the description is some text that the user would find helpful.

Our understanding of the semantics is:

   val is a statement about the story.

   type, parent, equiv and the description are statements about the 
   code.

   why, assign, conf, rel and when are statements about the 
   assignment of the code.

*  qual is a statement about ... umm ... don't know.  We understand 
   our requirement intuitively, but have trouble expressing it 
   formally.  Help!

I should also clarify that:

-  All of the above except for the code are optional.  So a minimal 
   syntax might be:
      <dc:subject val="cities:paris"/>
      <dc:creator val="comedians:groucho"/>
      <dc:type val="rhubarb:custard"/>

-  Our spec permits the provider to add an arbitrary amount of info, 
   which may be obtained by a recipient via the URI constructed from 
   the scheme:code QName.


Misha




------------- ---------------------------------------------------
        Visit our Internet site at http://www.reuters.com

To find out more about Reuters Products and Services visit http://www.reuters.com/productinfo 

Any views expressed in this message are those of  the  individual
sender,  except  where  the sender specifically states them to be
the views of Reuters Ltd.

Received on Friday, 10 June 2005 19:15:39 UTC