Re: please help define Web of Data

Giovanni Tummarello wrote:
> Hi,
> 
> I am writing here since i came across a few recent "foundational
> style" publications on the topic that give a definition  of "web of
> data" basically as SameAs of LOD - also specifying, in avoidance of
> doubt, that the LOD community started the web of data.
> 
> I wish to preserve and make clear the difference between "Web of Data"
> approaches such as Sindice.com (or anyone dealing with web markup
> really, Google, Facebook, Yahoo, Goodrelations ecommerce etc) and the
> technicalities that LOD considers fundamental requisites.
> 
> In our definition . (e.g. see the very beginning of the video at
> http://sig.ma ). the Web of Data,is the web made of pages that exposes
> machine processable content definited according to some metadata
> standard. So RDF, RDFa, Microformats, but also XML using notable
> schemas. Of course LOD is part of it.
> 
> In our view the *all these formats* do indeed serve the purpose of web
> scale data interoperability and aggregation. All these allow shared
> understandings thanks to shared vocabularies, so the differences are
> mostly syntactic and can easily be converted and integrated with
> similar, general tools.  [1] discusses a bit more the vision, though
> not specifically about this.
> 
> This is clearly outside LOD (it is indeed a vast superset). But i
> really apologize if we have used this term wrong so far.
> 
> I will appreciate and change the term if there is vaste feeling htat
> there would be no web of data without lod.
> Otherwise maybe those who mean LOD can call  it LOD?  :)
> 
> please advice.

Would be interested to see them define LOD clearly!

To me, the Web of Data comprises machine readable data (thus requiring a 
schema or media type) that's on the web (has a link to or from it, and 
is accessible via some well known protocol) - basically anything with 
about 2.5 stars or higher on the 5 star scale.

As for LOD, well I guess that's full 5 star data, data a machine can 
"think" about (reason over, apply some axioms etc), but until there's a 
uniform data format with a market penetration rate something like rss, 
atom or html, I don't think there's much point being too strict about 
it, after all a simple definition of a mapping from proprietary format x 
to rdf = rdf, to all extents and purposes.

Perhaps a more realistic and fairer definition is, if it's something 
that you guys at sig.ma or the openlink team can make your tools 
understand, and it's on the web, then it's part of the web of data :) 
That definition works for me!

kutgw,

nathan

Received on Tuesday, 19 April 2011 12:43:57 UTC