On 1/17/06, Zack Rosen <zack_at_civicspacelabs.org> wrote:
> > [Danny]
> > As Bruce, I'll concede that there is some merit in the arguments in
> > your blog post, but I think your title/conclusion ("Semantic web
> > research isn't working") is wide of the mark. Sure, some research is
> > relatively 'pure', away from practical implementation and commercial
> > applications. But not only is there plenty of down-to-earth research
> > going on in academia, there is a lot of research happening outside of
> > academia, with both corporate initiatives and individual hackers.
>
> Got urls?
You can find some at:
http://planetrdf.com/guide/
and
http://dannyayers.com/archives/category/virtual-world/semantic-web/sw-weekly/
> > Ok, there are qualitative differences in this point of view with SW
> > compared to WWW - a whole range of SW-compatible individual
> > implementations are possible from simple static text RDF/XML files
> > through to OWL inferencing models. In isolation they will only give a
> > taste. But any individual web page or service only gives a taste of
> > what the web has to offer.
>
> The WWW in it's conception offered unique critical functionality.
An arguable point. There were other approaches to the basic
functionality offered by the WWW - Gopher, WAIS etc. Clearly TimBL got
some significant details right, but that's more related to
implementation than functionality.
> Semantic web research projects so far have not come close to doing
> this.
I would suggest that the Semantic Web stack of specifications provides
critical functionality by allowing the web to generalise from
documents to data in a globally interoperable fashion.
Again, I don't see this happening until the gap between early
> adopters and the 'research community' is closed. The cheapest and
> most straightforward way to do this is to have research applications
> be developed on adopted open-source web application platforms in real-
> world use cases.
On that point I pretty much agree. I'd love to see more SemWeb
research connection with tools like Drupal, WordPress etc etc.
> >> I care deeply about the problem space. The issue is that it is
> >> simply too
> >> costly to 'jump on board' at the moment and I don't see that
> >> changing any
> >> time soon. Consider me an overly eager early adopter. I
> >> represent a number
> >> of organizations with reasonable development budgets that would be
> >> incredibly well served by semantic technologies but the tools are
> >> simply out
> >> of reach. Why is this?
> >>
> >
> > Costly in what sense? There is a large range of tools available open
> > source, most of them backed by active community support. The biggest
> > cost I'm aware of is in getting developers familiar with the
> > technologies (and managers to understand the benefits!). But that cost
> > is to be expected, virtually identical adoption issues were there in
> > the early days of the web.
>
> Got url's?
For which part? A lot of tools etc can be found via the links above
(and Google). Regarding adoption issues parallels, TimBL mentions them
in most of his interviews.
> > Your generalisation of "researchers" is pointing in a different
> > direction to the reality. A large proportion of Semantic Web R&D is
> > happening with an eye on developments elsewhere on the web. Because
> > the SW is designed to build on the current web. What TimBL said wasn't
> > "lets do it over", more "let's add these missing bits".
>
> The semantic web certainly is evolutionary but I can't see how it
> could be construed as anything but a huge leap from the assumptions
> and models of the web as-is. When created the semantic web will at
> the very least destroy and recreate business models, turn just about
> every web application into a scrap-heap, and significantly alter the
> ecology of web platform technologies.
I agree about there being significant changes to web ecology and (of
necessity) business models, but would suggest that current web
applications won't go to the scrap heap but evolve into Semantic Web
applications. For example, in the current setup the RDBMS-backed web
site is a common scenario. The front end of such systems are already
tied to the URI naming scheme of the Semantic Web. Because RDF can be
seen as a highly normalised relational model, connection from the back
end to the Semantic Web is possible without throwing anything away.
> > In fact there's clear evidence of this in the examples you quote,
> > because Semantic Web researchers *are* collaborating with the teams
> > pursuing these projects. If you look at the microformats list archives
> > you will see there has been significant input from people active in
> > the Semantic Web community.
> >
> > I personally have been working with the Structured Blogging folks on
> > Semantic Web compatibility - check the source of an SB entry, e.g.
> > http://opencontent.org/blog/archives/category/book-reviews/
> > - you will find the following :
> >
> > xmlns:data-view="http://www.w3.org/2003/g/data-view#"
> > data-view:interpreter="http://structuredblogging.org/subnode-to-rdf-
> > interpreter.xsl"
> >
> > Like most of the SB initiative this is experimental, but here a
> > mapping is provided from the SB information to the Semantic Web.
>
> I am sure collaboration is happing at an individual level but it
> certainly isn't happening at an institutional level. Why not?
There are different agendas. The point is well made elsewhere in this
thread re. researchers being motivated by getting their PhDs rather
than nifty implementations. Open source is its own reward (to quote
Tim Bray). There is a fair amount of institutional collaboration but I
suppose it tends to happen around less ad hoc initiatives than the
individual stuff (i.e. it's institutional ;-)
But is institutional level collaboration actually necessary? To what
extent has it occurred in the current web toolkit, e.g. Linux, Apache,
MySQL and P*?
> > Oh, go on, one last point - you say:
> > [[
> > The technologies are generally not useful unless they are adopted and
> > implemented on a large scale and people are not willing to invest in
> > implementing them unless they are useful.
> > ]]
> > Even without the Semantic Web, the technologies can be very useful for
> > a wide range of tasks, even on a very small scale.
RDF offers a very useful graph-shaped data model with a formal base,
and is worth considering for any application that uses
"semi-structured" data, whether they're standalone or networked. OWL
offers a systematic approach to knowledge representation, so if you
wanted e.g. to build a standalone expert system it would make a good
choice (Google "wine ontology" for an example). There are a bunch of
open source tool implementations for each (and of course they can be
used together).
Cheers,
Danny.
--
http://dannyayers.com
Received on Tue Jan 17 2006 - 11:31:03 EST