W3C home > Mailing lists > Public > public-lod@w3.org > September 2009

Re: how to consume linked data

From: Dan Brickley <danbri@danbri.org>
Date: Thu, 24 Sep 2009 14:44:51 +0200
Message-ID: <eb19f3360909240544gd2aea1by5be8b51c6634ed2b@mail.gmail.com>
To: Danny Ayers <danny.ayers@gmail.com>
Cc: Semantic Web <semantic-web@w3.org>, public-lod@w3.org
On Thu, Sep 24, 2009 at 9:59 AM, Danny Ayers <danny.ayers@gmail.com> wrote:
> The human reading online texts has a fair idea of what is and what
> isn't relevant, but how does this work for the Web of data? Should we
> have tools to just suck in any nearby triples, drop them into a model,
> assume that there's enough space for the irrelevant stuff, filter
> later?
>
> How do we do (in software) things like directed search without the human agent?
>
> I'm sure we can get to the point of - analogy -  looking stuff up in
> Wikipedia & picking relevant links, but we don't seem to have the user
> stories for the bits linked data enables. Or am I just
> imagination-challenged?

Good question.  I think part of the puzzle is in going back to some
document-centric ideas (RDF is generally centred around the stuff
documents describe).

For old slides on this and 'Hypertext RDF' from XML 2003 see

 http://www.oreillynet.com/xml/blog/2003/12/dan_brickleys_rdfsseealso_rdf.html
http://www.w3.org/2001/sw/Europe/talks/xml2003/Overview-3.html
all slides: http://www.w3.org/2001/sw/Europe/talks/xml2003/all.htm

Pre LOD, we were using a hypertext idiom that leaned more heavily on
rdfs:seeAlso to point to documents that had an RDF reading. I had a
slide on extensions:
[[

Extensions
We can do a lot with just rdfs:seeAlso, but RDF allows more. We can
give the type of the things rdfs:seeAlso relates:

<Person>
 <name>Dan Brickley</name>
 <rdfs:seeAlso>
   <x:Bibliography rdf:about="../stuffIwrote.rdf"/>
 </rdfs:seeAlso>
 <rdfs:seeAlso>
   <x:Resume rdf:about="../cv.rdf"/>
 </rdfs:seeAlso>
</Person>

This helps RDF crawlers be more discriminating, eg. by only traversing
links to bibliographies or other people, not bothering with CVs.

]]

This doc-typing idiom never got heavily used in FOAF, beyond the type
PersonalProfileDocument, which FOAF defines. Mostly we just linked
FOAF files together (initially with seeAlso and IFPs, lately using
URIs more explicitly).

I think there are many other reasons why characterising typical RDF
document patterns makes sense, related to the frustration of dealing
with documents when all you know is "they have triples in them". We
don't have good mechanisms for doing so yet, ie. for characterising
these higher level patterns. But various folk are heading in same
direction, some using SPARQL, others OWL or XForms, or DC Application
Profile definitions....

Without some hints about what we're pointing at with our links,
crawlers don't have much to go on. Merely knowing that the information
at the other end of the link is "more RDF", or that it describes a
thing of a certain type, might not be enough. There are a lot of
things you might want to know about a person, or a place, and at many
different levels of detail. For apps eg running in a mobile/handheld
environment, they can't afford to speculatively download everything..

cheers,

Dan
Received on Thursday, 24 September 2009 12:45:26 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 31 March 2013 14:24:23 UTC