W3C home > Mailing lists > Public > public-vocabs@w3.org > April 2014

Re: Socialnetworks of a person or organization

From: <martin.hepp@ebusiness-unibw.org>
Date: Fri, 11 Apr 2014 10:07:05 +0200
Cc: Kingsley Idehen <kidehen@openlinksw.com>, W3C Web Schemas Task Force <public-vocabs@w3.org>
Message-Id: <95787776-51D3-48F9-BDB6-EBDA9E0C55E7@ebusiness-unibw.org>
To: Dan Brickley <danbri@google.com>
Dan:
I am perfectly fine with your broad notion of linked data ;-) With "all linked data stuff", I meant primarily abandoning the use of two separate slash-based URIs for the type and its page.

The problem with many Linked Open Data recommendations is that they do e.g. not take into account the enormous amount of traffic caused by the use of HTTP prefetching, particularly by mobile clients and proxies in the infrastructure of mobile service providers. They often try to dereference any URI found in any href attribute, and if that returns an HTTP redirect, this consumes resources.

What I fear is are attempts to infiltrate the pragmatic, yet groundbreaking and far-reaching approach of schema.org with driving a narrow-minded notion of linked data. Of course, one can use schema.org in such settings (and that should be possible), but we should also be very reluctant with incorporating "narrow" LOD into the fibre of schema.org.

Martin


On 11 Apr 2014, at 09:46, Dan Brickley <danbri@google.com> wrote:

> On 11 April 2014 08:10, martin.hepp@ebusiness-unibw.org
> <martin.hepp@ebusiness-unibw.org> wrote:
> 
> An aside re
> 
>> I am planning to flatten the implementation in www.productontology.org so that there will be just one URI for the page and the type entity, essentially removing all linked data stuff in the near future.
> 
> <personal-soapbox>
> 
> It is a mistake to conflate "Linked Data" with this super-fragile
> deployment style.
> 
> The original Linked Data note
> (http://www.w3.org/DesignIssues/LinkedData.html) was essentially a
> response to the RDFWeb FOAF work at the time, and just said "let's try
> to label lots of these blank nodes with URIs that connect us to more
> RDF, instead of using rdfs:seeAlso". It is a pity if this insight has
> become enshrined in a mountain of complex, expensive and discouraging
> implementation detail, or a religious obsession with tieing every
> single node in every single RDF graph to an URL that publishes RDF.
> 
>> From Tim's original note:
> 
> """<#i> foaf:knows [
>       foaf:mbox_sha1sum "2738167846123764823647";  # @@ dummy
>  rdfs:seeAslo <http://example.com/foaf/joe> ].
> 
> This linking system was very successful, forming a  growing social
> network, and dominating, in 2006, the linked data available on the
> web.
> 
> However, the system has the snag that it does not give URIs to people,
> and so basic links to them cannot be made.
> 
> I  recommend (e.g in weblogs on Links on the Semantic Web , Give
> yourself a URI, and and Backward and Forward links in RDF just as
> important) that those making a FOAF file give themselves a URI as well
> as using the FOAF convention.     Similarly, when you refer to a FOAF
> file which gives  a URI to a person, use it in your reference to that
> person, so that clients which just use URIs and don't know about the
> FOAF convention can follow the link."""
> 
> I consider schema.org (yes, even in Microdata) to be quite reasonable
> Linked Data in this tradition. It is absolutely useful to have as many
> identifiers as possible, but we shouldn't allow it to become an
> obsession that overrides all other concerns e.g. webmaster usability.
> Linked Data as a slogan took off in large part because of frustration
> that "Semantic Web" had been taken over as slogan by people who cared
> more for scholarly research than for the Web. It is a great pity if
> the phrase "Linked Data" gets wasted in a similar fashion. IMHO
> Schema.org (and FOAF) are perfectly fine applications of the Linked
> Data idea. Nowhere in the sacred texts ("Use URIs as names for
> things", "Use HTTP URIs so that people can look up those names.", ...)
> does it say "In every document, for every entity.", nor "publish every
> relationship"; sometimes less is more. People worry too much. What
> we're creating here are languages that allow everyone on the Web to
> describe stuff in various levels of detail; we should let go of the
> urge to be so proscriptive about what exactly they publish (uris,
> string, blank nodes, etc.).
> 
> </personal-soapbox>
> 
> Dan
> 
Received on Friday, 11 April 2014 08:07:31 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:29:39 UTC