W3C home > Mailing lists > Public > semantic-web@w3.org > March 2005

Re: Announcement: RDF Content Labels

From: Benjamin Nowack <bnowack@appmosphere.com>
Date: Sat, 26 Mar 2005 17:17:32 +0100
To: "Phil Archer" <phil.archer@icra.org>
Cc: semantic-web@w3.org
Message-ID: <PM-EH.20050326171732.824D8.2.1D@192.168.27.2>


Hi Phil,

just had a quick glance at the content labels docs,
looks very interesting. I may be able to send more
detailed comments when I find the time to read the
whole stuff, just one thing I spotted:

- label:unionOf/intersectionOf are described as
  properties, but I think the way they are used
  inside an rdf collection makes them a class,
  e.g.:
[[
   <label:Ruleset>
      <label:rules rdf:parseType="Collection">
         <label:unionOf rdf:ID="rule1">
            <label:hasURI>ads</label:hasURI>
            <label:hasURI>banners</label:hasURI>
            <label:hasLabel rdf:resource="#label_2" />
         </label:unionOf>
         <label:intersectionOf rdf:ID="rule2">
            <label:hasURI>special-add-server</label:hasURI>
            <label:hasURI>customer_info</label:hasURI>
            <label:hasLabel rdf:resource="#label_2" />
         </label:intersectionOf>
      </label:rules>
   </label:Ruleset>
]]
would ceate the (pseudo-)triples
[[
   label:Ruleset label:rules _:list1
   _:list1 rdf:first #rule1
   #rule1 rdf:type label:unionOf
   ...
   _:list1 rdf:rest _:list2
   _:list2 rdf:first #rule2
   #rule2 rdf:type label:intersectionOf
   ...
   _:list2 rdf:rest rdf:nil
]]

not sure how to model this though, maybe just
[[
   <label:Ruleset>
      <label:rules rdf:parseType="Collection">
         <label:UnionRule rdf:ID="rule1">
            <label:hasURI>ads</label:hasURI>
            <label:hasURI>banners</label:hasURI>
            <label:hasLabel rdf:resource="#label_2" />
         </label:UnionRule>
         <label:IntersectionRule rdf:ID="rule2">
            <label:hasURI>special-add-server</label:hasURI>
            <label:hasURI>customer_info</label:hasURI>
            <label:hasLabel rdf:resource="#label_2" />
         </label:IntersectionRule>
      </label:rules>
   </label:Ruleset>
]]
to say that rule1 needs one condition to match,
and rule2 all conditions in order to apply another
label.

(Maybe changing "hasLabel" to "applyLabel" could
make the difference between conditions (matching URI)
and effects (change label) more obvious, but of course
that's only what a non-machine would say.. ;)

hth,
benjamin

--
Benjamin Nowack

Kruppstr. 100
45145 Essen, Germany
http://www.bnode.org/


On 23.03.2005 16:35:39, Phil Archer wrote:
>
>Dear all,
>
>Those at the second day of the SWIG meeting in Boston recently have had a 
>preview of this. Various people and organisations have been working on a 
>method of grouping URIs together and linking them to a common description, 
>what we call a content label. This is, I know, potentially confusing since 
>we're not talking about rdfs:label.
>
>ICRA's use case is that this can supersede PICS in offering a way for 
>content providers to label their content in terms of the various things 
>parents in different parts of the world don't want their children to see 
>(types of nudity, sexual material, violence, language, racism etc.). There 
>are others however - DC and Creative Commons information, for instance, will 
>generally apply to more than one of your resources. The mobile industry is 
>looking at a "Mobile OK" label for resources that work well on mobile 
>devices.
>
>The recent thread on this interest group between Benjamin Nowak and Stephen 
>Rhoads (SemWeb Non-Starter -- Distributed URI Discovery) might perhaps be 
>relevant here too since our aim is to make it possible to parse an RDF 
>instance that contains descriptions for a whole bunch of URIs. Our solution 
>is to create a simple ruleset that takes a URI as an input and gives a 
>description as output. In essence, the system amounts to:
>
>1. If you're on any of the following hosts, the default content label is 
>#here.
>2. There now follows an ordered sequence of regular expressions for you to 
>match against your target URI. As soon as you get a match, take the 
>associated label.
>3. If no match, thre default description applies.
>
>We offer machanisms that distinguish between a content label as a class that 
>has descriptive properties, a content label that delivers management 
>information (like DC and CC stuff), and a class that serves a description in 
>itself, such as "PG-13".
>
>Use cases and general information is at [1] with a description of the schema 
>itself at [2]. I have applied the system to ICRA's use case and created what 
>this group might call an RDF generator at [3] that allows you to "label your 
>website" and a basic tool to test those labels at [4]. The latter is very 
>much under development and will make full use of SPARQL in due course and 
>may do other fancy things - as ever, it's all about time and priorities!
>
>Incidentally, Piggy Bank [5] shows up the content labels nicely, but, being 
>unaware of the rule methods we've proposed, shows all available labels 
>without knowing which label applies to the particular page you're on.
>
>Comments more than welcome.
>
>Thank you
>
>Phil Archer
>www.icra.org
>
>[1} http://www.w3.org/2004/12/q/doc/rdf-contentlabels.html
>[2] http://www.w3.org/2004/12/q/doc/content-labels-schema.htm
>[3] http://www.icra.org/RDF/label/
>[4] http://www.icra.org/RDF/label/tester/
>[5] http://simile.mit.edu/piggy-bank/ 
>
Received on Saturday, 26 March 2005 16:17:55 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:41:45 UTC