Re: call to arms

The original question was why has not adoption been broader at this point,
asked in different forms. What are the barriers?
There is no one answer. The answer is not solely in the way the technologies
are presented either by people or approached through tools or any other
possibility including the nature of the data.
Nor is the answer in what I am going to bring to your attention.
If we look at the adoption of foaf we can see how things can be rapidly
adopted.
But when I suggested exploring semantic technologies at Serco PLC as a
solution to some of their problems in surfacing information for businesslink
(on behalf of the UK government and HMRC) this was point blank refused.
It is reasonable to consider why this might be.
Defending their own position from a threat, cost, other factors?
Thinking about cost, is the cost of introducing foaf so much less than a
more comprehensive semantic solution?
Immediately the question is asked and I realise that I am not in possession
of any cost benefit analysis for reference. I don't think they exist yet.
Looking at the adoption of foaf I realise that probably in the case of its
adoption cost benefit analysis was not necessary, since the costs were very
low.
One of the reasons for the disjoint between research and business
implementation is that academic research is set up to produce papers that
present results, but not field implementations. Of course not.
Field implementations are expensive.
I have suggested that one reason for enthusiasm for a public semantic web is
to defend by example against data corruption for gain.
But another maybe that there is a lock in to this vision because other
possibilities are inaccessible?
Media companies such as the BBC and its rivals on one scale and google, on
another, are grappling with difficult problems such as semantic markup of
video. Niggling away at these problems solutions emerge and other web sites,
businesslink perhaps, will fall into line.
This happens at the point that not taking up the technology is more
expensive than its adoption.
With businesslink I expect it will be mark up of assets, where they are
stronger, rather than an approach to underlying core technology, where they
are week. (Although this begs the question of how they would support it, but
still ... )
Semantic technology is really too vast a subject area, just think of the
chasm between the semantics as information architecture and semantics in a
core application that I have just alluded to.

It is also incredibly interesting.
There are so many unanswered questions in my mind.
For instance, on the subject of the soundness of data, thinking in terms of
a semantic application invites the production of different data than that
traditionally stored, data about the application itself.
We do not know to what extent a traditional core application could be
converted into a semantic engine.
What data might it produce to feed its own functioning?
What overlap is there here between site usage statistics and other aspects
of the application?
What would be left out from the traditional design patterns, and what new
design patterns would be used?
Can a semantic application be developed that would have semantics as a first
class design pattern just as MVC is now? I am not at all sure what such an
application looks like architecturally.
On the level of code, I want an annotations framework that allows me to drop
in semantic descriptions which work like aspects, (actually that are
aspects, but not necessarily using aspectj, how about run time aspects? Or
traits in Scala?) because I think that the relationship between a semantic
class and its properties is as that between an object and object aspects,
those properties imply behaviours that aspects on an object would express.
Or should I rely on a script to tie things together, which is another
approach?

Many issues ...

Adam




On 18 April 2010 14:17, Dan Brickley <danbri@danbri.org> wrote:

> On Sun, Apr 18, 2010 at 1:11 PM, Danny Ayers <danny.ayers@gmail.com>
> wrote:
> > On 18 April 2010 12:54, Michael Schneider <schneid@fzi.de> wrote:
> >> Danny Ayers wrote:
> >>
> >>>when do I plant my tomatoes?
> >>
> >> We are in early Spring now. Tomatoes don't grow well in this period. At
> >> least not in the outside. Well, you can find them in the greenhouse, but
> >> that's probably not what you are looking for. So, I'm afraid, you have
> to be
> >> patient.
> >
> > Thank you Michael, but I wish to make you redundant. This box of
> > circuits in front of me should have told me that.
> >
> > Did you take into consideration that I live on this side of the
> > Garfagnana valley?
>
> When I think about linked information these days, I see three major
> flavours:
>
> * information in classic document form (analog stuff made of bits;
> human-oriented prose, video, imagery)
> * information in source-attributed RDF claims (aka Linked Data, quads, etc)
> * information in people's heads
>
> For me, the RDFWeb/FOAF story I think has always been about the 3-way
> relationship between these different equally important ways of
> learning about the world. Linked people *and* linked information.
>
> You can think of lots of aspects of SemWeb as positioned as edges of
> this simple triangle where the nodes are the categories above. RDF
> syntaxes, GRDDL for microformats, RDFa, Adobe XMP, ebook metadata,
> Dublin Core etc are often links between classic document forms and RDF
> quads. Sometimes RDF quads are more to summarise what the document
> says about the world; other times they are to help find it. Similarly,
> provenance, authorship and other people-describing RDF, also
> people-describing non-RDF information, can all help us to find whose
> *head* might have the right information. A YouTube video can capture
> something of a person's subjective knowledge of the world and put it
> out there in document form for others to find; tags and RDF stuff can
> help others find that video and either learn directly or get in touch.
> SemWeb people (all of us) can easily focus only on one of these forms
> of information, at the expense not only of the other two, but their
> rich interconnections. Machine-unfriendly video, images, audio or .xls
> files can still be very useful, and the 'RDF as metadata about files'
> use case is one we too easily neglect.
>
> > Did you take into consideration that I live on this side of the
> Garfagnana valley?
>
> In this case I think the answer is best found in the heads of your
> neighbours, rather than on the Web. How's your Italian coming along?
>
> cheers,
>
> Dan
>
>