Loading ...
Sorry, an error occurred while loading the content.

776RE: [newsml-g2] Properties of a concept

Expand Messages
  • Laurent LE MEUR
    Dec 23, 2008
    • 0 Attachment
      This is true that, for concept/entity descriptions, NewsML-G2 offers
      - related = primarily a "see also" relationship
      - facet = primarily a "is a" relationship (this place is a restaurant ...)
      - the possibility to use specific properties from other namespaces for other use cases (e.g. RDF properties)

      But the "rel" (relationship) attribute supported by both <related> and <facet> makes the choice more problematic because itcan change the semantics of the property.

      I would propose the following guideline :
      - use related when the value is a concept identified by a URI, part of a network of concepts in a knowledge base (be careful : this is NOT the way EventsML structure have been designed; and the explicit property person->affiliation->organization neither fits this pattern).
      - use facet for subclassing concepts (this location is a city ...)
      - use other properties for typed values, especially integers. This is your case, Paul.

      Laurent

      -----Message d'origine-----
      De : newsml-g2@yahoogroups.com [mailto:newsml-g2@yahoogroups.com] De la part de darko.gulija@...
      Envoyé : mardi 23 décembre 2008 01:41
      À : newsml-g2@yahoogroups.com
      Objet : Re: [newsml-g2] Properties of a concept

      Michael,

      First, we already have <capacity> property in POIdetails, so this would be a logical placeholder for number of seats for a cinema.

      Second, (on the philosophical level - if a question is in fact "how to express some plain characteristics of a concept that is an integer", and not exactly number of seats), I would rather use <facet> than <related>.

      Facet is defined as "An intrinsic property of a concept", whilst <related> is defined as "An identifier of a related concept, where the relationship is different from elements sameAs, broader, or narrower."

      Number of seats is much more property of a cinema than relation between the concept of cinema and the concept of "135".

      We defined that facet, by default, expresses "isA" relation, but we also provided "@rel" attribute to express some other characteristics (e.g. "hasA" - remember "is a bar" against "has a bar" discussion).

      So, facet like
      <facet rel="myfacets:noOfSeats" literal="5">
      seems quite logical (e.g for a car - which is not a POI, so it does not have <capacity>).

      Another question is how to express the value that is an integer, and not concept itself.

      In POIdetails we defined discrete properties when the value was not the concept, but integer or some other datatypes (e.g. date in born/died for a person etc).

      So, IMO the options are:
      1. Use facet, and treat literal as integer
      2. Define discrete property in your own namespace
      3. Propose extension of the IPTC datatype with new property, if characteristics is universal enough.

      Darko Gulija
      Savjetnik za razvoj IT sustava
      Counsellor for IT strategy and development
      tel: +385 1 48 08 800
      fax: +385 1 48 08 820


      ----- Original Message -----
      From: "Michael Steidl (IPTC)" <mdirector@...>
      Date: Monday, December 22, 2008 6:42 pm
      Subject: Re: [newsml-g2] Properties of a concept
      To: Paul Harman <paul.harman@...>, newsml-g2@yahoogroups.com


      > On 22 Dec 2008 at 17:04 Paul Harman wrote:
      >
      > > Hi,
      > >
      > > I'm examining adopting G2 as a common interchange format between PA
      > > content systems. Regarding typical article-based content it's a fairly
      > > obvious choice of standard. However I also want to exchange data 9as
      > > related to articles) - cinema listings, TV guides, sports results,
      > > election results, school league tables.
      > >
      > > My particular query is to do with how I can extend conceptItems to
      > add
      > > new kinds of facts - for example, how many seats a particular cinema
      > > has. The concept of 'number of seats' could come from a controlled
      > > vocabulary of 'properties of a cinema', but the /value/ for number
      > of
      > > seats cannot practically come from a controlled vocabulary. But that's
      > > not my understanding of how Qcodes work... Shouldn't every Qcode be
      > > 'resolvable' as an item in it's own right?
      >
      > Paul, I just recently discussed something like that with dpa
      > regarding EventsML-G2:
      >
      > - for fine grained descriptions of a concept use the <related>
      > property (be aware: the <facet>
      > is intended to refine the type of the concept, e.g. if the type is
      > "person" you may add the facet
      > "male" to it.)
      >
      > - a <related> property provides:
      > -- a @rel to express the kind of relationship, e.g. "numberOfSeats"
      > -- a @qcode or a @literal for the value
      >
      > This way you could add e.g.
      >
      > <related rel="poiProps:noOfSeats" literal="83" />
      >
      > <related rel="poiProps:operator" qcode="pacompanies:UCI" />
      >
      > (What I discussed with dpa: this way you can also add the CEO of a
      > company - and more)
      >
      > >
      > > Is this where the G2 standard 'ends' and I need to invoke my own XML
      > > fragments via the extensibility options - or, like NewsML 1.x, is there
      > > a generic 'Property' element that I've overlooked?
      >
      > I still consider the G2 design as covering many, many requirements,
      > this way you can avoid
      > introducing new properties, you only have to add the corresponding
      > taxonomies for
      > relationships.
      >
      > Michael
      >
      > >
      > > Paul
      > >
      > > ------------------------------------
      > >
      > > Any member of this IPTC moderated Yahoo group must comply with the
      > > Intellectual Property Policy of the IPTC, available at
      > > http://www.iptc.org/goto/ipp. Any posting is assumed to be submitted
      > > under the conditions of this IPTC IP Policy.
      > > Yahoo! Groups Links
      > >
      > >
      > >
      > >
      >
      >
      > ==================================================
      > Sent by:
      > Michael Steidl
      > Managing Director of the IPTC <mdirector@...>
      > International Press Telecommunications Council
      > "Information Technology for News"
      > Visit us on the web at http://www.iptc.org
      >
      >

      ------------------------------------

      Any member of this IPTC moderated Yahoo group must comply with the Intellectual Property Policy of the IPTC, available at http://www.iptc.org/goto/ipp. Any posting is assumed to be submitted under the conditions of this IPTC IP Policy.
      Yahoo! Groups Links



      This e-mail, and any file transmitted with it, is confidential and intended solely for the use of the individual or entity to whom it is addressed. If you have received this email in error, please contact the sender and delete the email from your system. If you are not the named addressee you should not disseminate, distribute or copy this email.

      For more information on Agence France-Presse, please visit our web site at http://www.afp.com
    • Show all 9 messages in this topic