Re: ANN: Topics, Tags and Ontologies - [was IPTC releases first draft of the rNews standard for embedding metadata in online news]

I agree that if you stay in your local world of tagging, and noone will ever
be inclined to try and link to other ontologies, then there is no chance of
a problem.  It may work in this situation. However, it seems odd to me that
you would go to the trouble of publishing this in a public standard notation
if you only cared about your closed world.

It really depends on what you mean by a tag.  Your ontology allows an
organization to be a tag and a person to be a tag.  What you really mean is
that you need a way to say that an article is in some way related to that
organization or person. You don't have to have a class called Tag to do
that, all you need is the property (I forget what you called it - maybe
isTaggedBy) to relate the article to the organization.

Given that anything at all can be used as a Tag, Tag has no more semantics
really, than owl:Thing.  An article can refer to just about anything.

One way to view a tag is that it is a topic. Something that an article is
about.   An organization is not a topic per se, nor is  a person.   Yet,
oddly, a person can also be the topic of a book (e.g a biography).  A topic
does not have blood pressure, a person does. A tag does not have blood
pressure either.  To do this "properly" gets messy, you have to distinguish
PersonAsTopic from Person which in most cases is unnecessary.

If you are 100% certain that no problems will ever arise if your ontology
infers that a tag has blood pressure, then you will not have any problems of
the sort that concern me.

I hope it works out for you.

Michael

On Wed, Apr 6, 2011 at 5:55 AM, Bernard Vatant
<bernard.vatant@mondeca.com>wrote:

> I'm with Richard on this
>
> rNews is a functional ontology, which means intended for a functional
> purpose : tag news. In this context having Organization, Person and Location
> subclasses of rnews:Tag makes perfect sense.
> If you declare in your ontology ex:Person  owl:disjointWith rnews:Tag it
> means that you don't want to tag news with your instances of ex:Person. So
> what? That's what you want. I don't see any problem with that. I'm sure no
> big problem would come from something like rnews:Person  rdfs:subClassOf
> foaf:Person.
> foaf:Person does not make any strong assumption on what a Person is, that's
> why it's so easily reused. A rnews:Person is a foaf:Person who happens to be
> used for tagging news. Certainly that will make some reasoners crash at some
> point if they try to make consistent too many things at a time.
> In a more restricted scope of FYN applications, it will be very useful. For
> example I have a data base of people using foaf:Person I can scrap rNews to
> populate it. I don't care if a rnews:Person is a Tag or not, but am happy to
> identify a foaf:Person when I meet her in the news.
>
> Interpret what makes sense to you, ignore the rest.
>
> Best
>
> Bernard
>
>
> 2011/4/6 Richard Cyganiak <richard@cyganiak.de>
>
>> Hi Michael,
>>
>> On 6 Apr 2011, at 03:48, Michael F Uschold wrote:
>> > I had a look at the ontology and discovered that a) it does not load
>> into Protégé 4.1.  and b)    An organization and a location and a person are
>> all tags.  This seems likely to cause semantic problems if you want to
>> interoperate with any other data out there.
>> > For example, someone is likely to say
>> http://iptc.org/std/rNews/2011-02-02#Organization is owl:sameAs the URI
>> for organization from say DBpedia or the CIA World Fact Book.  Now all these
>> are going to be inferred to be of type
>> http://iptc.org/std/rNews/2011-02-02#Tag.   This will give an
>> inconsistent ontology when someone does the sensible things and says a
>> location and an person are disjoint from tags.
>> >
>> > A tag is an informational indexing thing. An organization is a legal
>> entity, a location is something else again.
>>
>> And I thought a tag is a small, flat, metal thing on a dog collar…
>>
>> rnews:Tag doesn't have to be the same as michael:Tag. That's why we name
>> things with URIs.
>>
>> If it is useful *for some application* to declare rnews:Organization
>> owl:equivalentClass ex:Organization in their own local application-specific
>> ontology mapping, then ex:Organization and rnews:Tag are not disjoint *in
>> that mapping*. That's fine. All mappings are wrong, but some are useful.
>>
>> The Web will never be globally consistent and you know that.
>>
>> Best,
>> Richard
>>
>>
>>
>> >
>> > Michael
>> >
>> >
>> > <!-- http://iptc.org/std/rNews/2011-02-02#Organization
>> >  -->
>> >
>> >     <owl:Class rdf:about="#Organization">
>> >         <rdfs:subClassOf rdf:resource="#Party"/>
>> >     </owl:Class>
>> >
>> >
>> >
>> >     <!--
>> > http://iptc.org/std/rNews/2011-02-02#Party
>> >  -->
>> >
>> >     <owl:Class rdf:about="#Party">
>> >         <rdfs:subClassOf rdf:resource="#Tag"/>
>> >     </owl:Class>
>> >
>> >
>> > semantic-web@w3.org
>> >
>> > ---------- Forwarded message ----------
>> > From: Raphaël Troncy <raphael.troncy@eurecom.fr>
>> > Date: 2011/4/5
>> > Subject: ANN: IPTC releases first draft of the rNews standard for
>> embedding metadata in online news
>> > To: Semantic Web <semantic-web@w3.org>
>> >
>> >
>> > Dear all,
>> >
>> > Short Story:
>> > ------------
>> >  - IPTC has released today the rNews standard draft, see
>> http://dev.iptc.org/rNews
>> >  - rNews aims at marking news using RDFa. The standard will be compliant
>> with NewsMLG2, IPTC’s XML format for multimedia news packages
>> >  - IPTC considers incorporating hNews into its set of standards
>> >  - rNews will be machine readable, an OWL/RDF version of the vocabulary
>> is being cooked
>> >
>> > IPTC is requesting _NOW_ feedback from the community, via its dedicated
>> forum, http://dev.iptc.org/Forum-1. The IPTC will consider any feedback
>> posted until 30 April 2011 as input to the final standardization process.
>> >
>> > Long Story:
>> > -----------
>> > Publishers large and small struggle to make sure that search engines and
>> social media sites find their stories and refer to them appropriately.
>> >
>> > They want to provide highly targeted adverts while dealing with users
>> who are opposed to the privacy implication of sharing the personal data
>> necessary to accomplish that.
>> >
>> > How can they build web pages with news stories where the components of
>> the story are machine-readable, as well as human readable? The rich content
>> metadata available in a publisher’s internal CMS is often lost, or
>> re-engineered in an attempt to optimise search results.
>> >
>> > The IPTC has taken a step to solving this problem with the release of
>> the first draft of the rNews standard in 5 April 2011.
>> >
>> > The IPTC aims to revamp the process for marking up online news articles
>> so that search engines and social networking sites can more successfully
>> pull relevant pieces of data from them. IPTC seeks to develop rNews, a
>> recently proposed practice for marking news using RDFa, which allows HTML
>> authors to tag webpage content with indicators that browsers and tools can
>> interpret. The standards will be compliant with NewsMLG2, IPTC’s XML format
>> for multimedia news packages.
>> >
>> > IPTC’s Semantic Web Working Group is leading the charge to craft these
>> standards. The group is mapping out the specifications essential to rNews
>> for IPTC, and will offer examples of the standards in use, best practices
>> for implementation, and online tools to ease the transition to the new
>> standards. This work is set to conclude in the summer of 2011, and will
>> better enable search engines to cull NewsML-G2 data from the Internet, ready
>> for Semantic Web and Linked Data mining technologies.
>> >
>> > Details of rNews are available at http://dev.iptc.org/rNews.
>> >
>> > More importantly, what you think about it can be recorded on the public
>> rNews forum (http://dev.iptc.org/Forum-1) . The IPTC will consider any
>> feedback posted until 30 April 2011 as input to the final standardization
>> process.
>> >
>> > The IPTC considers incorporating hNews, the existing microformat for the
>> semantic markup of news content, into its set of standards.
>> > At its most basic level, both hNews and rNews are about expressing news
>> metadata in HTML.
>> > hNews is a microformat using HTML markup conventions, whereas rNews is a
>> method to record RDF assertions (triples) within HTML, but keeps
>> vocabularies and expression mechanism separate.
>> >
>> > Best regards.
>> >
>> >  Raphaël Troncy
>> >  on behalf of the IPTC SemWeb working group
>> >
>> > --
>> > Raphaël Troncy
>> > EURECOM, Multimedia Communications Department
>> > 2229, route des Crêtes, 06560 Sophia Antipolis, France.
>> > e-mail: raphael.troncy@eurecom.fr & raphael.troncy@gmail.com
>> > Tel: +33 (0)4 - 9300 8242
>> > Fax: +33 (0)4 - 9000 8200
>> > Web: http://www.eurecom.fr/~troncy/
>> >
>> >
>> >
>> >
>> > --
>> > Michael Uschold, PhD
>> >    Senior Ontology Consultant, Semantic Arts
>> >    LinkedIn: http://tr.im/limfu
>> >    Skype, Twitter: UscholdM
>> >
>>
>>
>>
>
>
> --
> Bernard Vatant
> Senior Consultant
> Vocabulary & Data Engineering
> Tel:       +33 (0) 971 488 459
> Mail:     bernard.vatant@mondeca.com
> ----------------------------------------------------
> Mondeca
> 3, cité Nollez 75018 Paris France
> Web:    http://www.mondeca.com
> Blog:    http://mondeca.wordpress.com
> ----------------------------------------------------
>



-- 
Michael Uschold, PhD
   Senior Ontology Consultant, Semantic Arts
   LinkedIn: http://tr.im/limfu
   Skype, Twitter: UscholdM