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

Re: Making RDF / LinkedData trivially browseable - thoughts?

From: Michael Hausenblas <michael.hausenblas@deri.org>
Date: Sat, 04 Apr 2009 09:00:41 +0100
To: Hugh Glaser <hg@ecs.soton.ac.uk>, Kjetil Kjernsmo <kjetil@kjernsmo.net>, Daniel O'Connor <daniel.oconnor@gmail.com>
CC: Linked Data community <public-lod@w3.org>
Message-ID: <C5FCD239.36D8%michael.hausenblas@deri.org>

Daniel, All,

> I am rather uncomfortable with even the idea that users might "browse the
> semantic web".

I sort of second Hugh, here. Various reasons, really, but basically Web of
Data to me mainly means doing something in the background. A human might
ideally not even notice that she is 'in' the Web of Data, just benefiting
from it.

How?

Dunno. We gotta find out, but in a problem-pull rather than technology-push
fashion, I believe. Look at stuff you have been doing recently and ask
yourself (similar to  BBC [1] maybe?): what part of my app can be done
better/more efficient/with less effort/etc.

I tried to put together some initial thoughts that might help as well [2]
(pre-print of my upcoming IEEE Internet Computing journal article).

I also second Kjetil when he says: forget about the XML tool chain. Luckily
the Semantic Web stack has changed some time ago and now acknowledges that
very fact (RDF is *not* build on top of XML; for a long time a very serious
marketing problem, IMHO).

On the other hand we see technologies such as RDFa, potentially being the
next big step in the linked data world; a very natural way [3] to deploy
linked data, if you ask me.

Cheers,
      Michael

[1] http://www.bbc.co.uk/blogs/radiolabs/2009/01/how_we_make_websites.shtml
[2] http://sw-app.org/pub/exploit-lod-webapps-IEEEIC-preprint.pdf
[3] 
http://events.linkeddata.org/ldow2008/papers/06-halb-raimond-building-linked
-data.pdf

-- 
Dr. Michael Hausenblas
DERI - Digital Enterprise Research Institute
National University of Ireland, Lower Dangan,
Galway, Ireland, Europe
Tel. +353 91 495730
http://sw-app.org/about.html
http://webofdata.wordpress.com/


> From: Hugh Glaser <hg@ecs.soton.ac.uk>
> Date: Sat, 4 Apr 2009 00:32:34 +0100
> To: Kjetil Kjernsmo <kjetil@kjernsmo.net>, Linked Data community
> <public-lod@w3.org>
> Cc: Daniel O'Connor <daniel.oconnor@gmail.com>
> Subject: Re: Making RDF / LinkedData trivially browseable - thoughts?
> Resent-From: Linked Data community <public-lod@w3.org>
> Resent-Date: Fri, 03 Apr 2009 23:34:07 +0000
> 
> Hi,
> OK Kjetil, I clearly need to respond to your injunction to respond to Daniel.
> Perhaps I avoided answering because it would feel a bit negative - not sure.
> I am rather uncomfortable with even the idea that users might "browse the
> semantic web".
> I know that is not what you are suggesting, Daniel, but you are also implying
> that the way a developer might get the sense they are gaining something with
> this Linked Data (LD) stuff is by clicking from site to site.
> Unfortunately (for us?!), the real advantage of LD is meant to be more subtle.
> 
> I would split your "immediate payoff" question into two parts:
> 
>  1.  Consumer
> 
> As a Consumer, an example would be that my system had a URI for a person (from
> my domain), and because it also knew the dbpedia URI for that person, it was
> able to publish the institution name for the person if dbpedia had one.
> Another example would be to be able to identify where they were located,
> because of inferring from sites such as dbpedia where the institution they
> come from is located.
> 
>  1.  Producer
> 
> As a Producer, having published this LD, I would find that there were
> applications around that took cognisance of this LD, so that the next time I
> went and looked at that data, it would (in some way) include the data that I
> was publishing.
> 
> Both of these require quite sophisticated applications that might for example
> resolve URIs and perhaps use the information by some sort of fresnel
> rendering, without being tied to particular ontologies, and these are not
> widespread - hence your question, I think.
> However, I think that the benefits of Consumerism can be quite well
> demonstrated, as long as it is easy to find the coreferent URIs to make the
> linkage.
> 
> So in summary:
> Tell your "regular web developer guy" that all he needs are the URIs for the
> thing he is concerned with, so he can resolve them, and then he can do a
> "Mash-up/Mesh-up" with the data he gets back to offer lots more data to his
> customers, from a very wide range of sites.
> 
> Hope that helps.
> Hugh
> 
> On 03/04/2009 22:38, "Kjetil Kjernsmo" <kjetil@kjernsmo.net> wrote:
> 
> On Wednesday 25 March 2009, Daniel O'Connor wrote:
>> Hey all,
> 
> Hey Daniel!
> 
> I'm a bit embarrassed that your questions have gone unanswered, as I
> think they are very relevant and important.
> 
>> So, here's the scenario. I'm a regular web developer guy and I've
>> heard about linked data. I know just enough about webservices and
>> rest and xml and that sort of thing, and I'm sold on the big picture
>> of the semantic web; and I've now just come across linked data.
> 
> Cool! Welcome! I think we really need people like you coming on board!
> 
>> Q: What's available to help me know "when I'm doing it right"; and
>> what's available to make it feel like there's an immediate payoff?
> 
> So, these are very good, but quite difficult questions to answer right
> now, but I think they are important questions to answer. I kicked off
> the Community Projects within the Semantic Web Education and Outreach
> interest group, which lead to the LOD back in the day, partly because I
> felt that questions like these too often went unanswered. Now, we have
> tons of data and relatively well established best practices, but I
> still feel that the community needs to come together and answer these
> questions, and to be a bit provocative, I suspect that the reason your
> email has gone unanswered is that we aren't there yet. So, folks,
> answer the questions! :-)
> 
>> 
>> So far, the simplest answer I have to that question is "whack a
>> simple xsl ontop of it so the RDF gets rendered as
>> not-very-pretty-but-hey-i-can-click-links html".
>> 
>> Criteria for success: I can click from one half of my data set to the
>> other, then end up at dbpedia, and then click off to somewhere else;
>> and it feels like a unpretty normal web.
>> 
>> 
>> 
>> So, to that end, I've whipped up this really quick and dirty
>> stylesheet to basically do that. See
>> http://pastebin.com/pastebin.php?dl=f289d4f5c
>> 
>> Preview it by sticking this into your xml (and obviously host your
>> own decent copy if you are using it in The Real World)
>> 
>> <?xml-stylesheet type="text/xsl"
>> href="http://pastebin.com/pastebin.php?dl=f289d4f5c" ?>
>> 
>> What do you guys think of this approach?
> 
> I'm actually not quite sure exactly what you're trying to achieve. There
> are a bunch of generic data browsers, Tabulator, OpenLink Data Explorer
> (FF extensions), Disco Hyperdata Browser, etc. So, it seems to me those
> could be used for browsing data.
> 
> I've been writing huge amounts of XSLT for my apps (basically doing XSLT
> on a constrained RDF/XML tree), and it gets really, really ugly pretty
> quick. So, I'd discourage people from going down that path. In fact, it
> seems like we need to abandon the entire XML toolchain for most of the
> stuff we do (well, eventually, you could use some HTML, where it has
> relevance, but other than that, throw it out).
> 
> So, there has been quite some talk about what to do, where Fresnel
> http://www.w3.org/2005/04/fresnel-info/ may be the thing that has the
> most traction. I think it looks like a massive overkill for most
> regular web developers, and for my purpose too, so I started to think
> about a RDFa templating language:
> http://www.kjetil.kjernsmo.net/software/rat/
> I haven't gotten any further on it, and it needs more thinking. And
> since I'm more of an RDF geek than an X?(HT)?ML geek, it probably could
> use some thinking from a regular web developer too :-)
> 
> Another thing I'd like to do is use SPARQL DESCRIBE queries to just
> generate the data and the labels and stuff you'd need to present stuff
> to a human user, and I put some of those ideas down here:
> http://www.w3.org/2009/sparql/wiki/Feature:ControlOfDescribeQueries#Detailed_u
> se_case:_Hardware_components_search
> That also needs a bit more thinking, and the standards aren't even
> there, so there is still work to do here. :-)
> 
> Cheers,
> 
> Kjetil
> --
> Kjetil Kjernsmo
> Programmer / Astrophysicist / Ski-orienteer / Orienteer / Mountaineer
> kjetil@kjernsmo.net
> Homepage: http://www.kjetil.kjernsmo.net/     OpenPGP KeyID: 6A6A0BBC
> 
> 
> 
Received on Saturday, 4 April 2009 08:01:24 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 31 March 2013 14:24:20 UTC