Re: "State of the Semantic Web" - personal opinions?

Danny, Paul and all

Hopefully useful discussion, thanks for opening it up - I think we all have 
different views of what semantic web is
I take the opportunity to sell our bit of perspective

(note our forthcoming workshop below, and please let us know if someone 
wants to jump on board and give us a hand here
http://sites.google.com/site/humanfactorsandsemanticweb/Home)

I am very interested in the points Paul T makes below. In a way its kinda 
true, what you say, that at the moment it looks a bit silly
in another way it's kinda not true. Let me explain what I see.
What I call 'semantic web' is still a bit of a vision (TBL vision perhaps, 
expanded)

That is: I can log onto the web, and using a smart and user friendly 
interface and query all the data that exists online in any way I wish, 
including the relationships between data sets. In addition, I can search for 
qualitative/fuzzy terms.

For example: how many cities of more than 100.000 inhabitants exist in 
Europe?
 (and reason with the result, for example: for each city over 100k, 
calculate the ration nr of vehicles per person/CO2 density etc)
or
How many companies are hiring people with my skillset in the UK? (and 
automatically parse the emails and send my cv to all)
or
where can I find the nearest (relative to) producer of xyz goods? (produce a 
report ranked based on different criteria incl prices)
or even more simply
How many people on this list are intersted in UI for semweb? of these, how 
many have replied to my emails, how many I have met already and
how many are looking for project partners? (sad I cant do that yet)

At the moment, the following problems exist (probably more, but for 
simplicity)

1. not all data is on the web
2. not all data on the web is valid/updated
3. not all data which is valid and updated is expressed in a format that can 
be intelligently queried by a browser (that I can download and run)
4. the current data representation and querying technologies (rdf. owl and 
sparql) are not necessarily the most efficient way to achieve such 
capabilities
5. even semantic browsers do not offer inteligent reasoning capabilities


So I see two sides to the problem -

a) one is the data representation. How can I, mere mortal 'business entity' 
(presumably I hold data and publish it on the web) make my data available 
and suitable for intelligent querying? Where is the website that can digest 
my data and produce god valid xml/rdf (external) representation (a la calais 
type, but even more usable).

Lets not forget that organizations are just coming to terms now (policy, 
technology, budgeting, human resourcing etc) with putting their stuff on the 
web
People are still very cautious and uncertain about web publishing, 
especially when it comes to business data.

Semantic web technology is currently a geeky affair. But can you see the BI 
(business intelligence) potential in there? If you had all the data
available on the web, validated and ready to be queried at your fingertips, 
would you not be able to save time and money in your decision making?

b) the other is building these intuitive  intuitive/usable/robust search 
interfaces which can be queried using formal, semi-formal and natural 
languages
(possibly combined)


As for the requirements of the aircraft, I am not sure a semweb application 
would be really  be the answer in the first place

Requirements for aircraft only minimally depend on the data available on the 
open web (unless you can dynamically and selectively parse and aggregate 
from the web an ontology, but thats really a bit far ahead), unless you are 
thinking of deployment over intranet. Even so, I think what you are after 
there is a 'knowledge engineering' function (make sure the data/process 
structure is aligned and synchronised) not sure if I see that as a core 
semweb functionality, although I am sure it can be used to support such a 
capability.

Just another two cents (I am in the UI camp, ready to work on that front)

Paola Di Maio




----- Original Message ----- 
From: "Paul Tyson" <phtyson@sbcglobal.net>
To: <semantic-web@w3.org>
Sent: Monday, June 30, 2008 11:11 PM
Subject: Re: "State of the Semantic Web" - personal opinions?


>
> What a beautiful invitation, Danny!
>
> From my sliver view inside a typical American corporation, semantic web 
> has a long road ahead.  No business man will pay $1.00 for a system that 
> will tell him that his mother's sister is his aunt, which is about all 
> that SemWeb101 shows you.  But show him an application that will tell him 
> if his billion-dollar aircraft design meets all 
> rehttp://sites.google.com/site/humanfactorsandsemanticweb/Homequirements, 
> and you'll be in for some money.
>
> As with all standards that enable information owners to control their 
> information, the semantic web profit model is elusive.  When tools for 
> knowledge representation and exchange again become commodities (as they 
> were for most of the Gutenberg age), the market will encourage people to 
> compete on their ability to think and provide value, instead of just 
> charging license fees for locking up your enterprise data.  On the other 
> hand, since semantic web content is, ultimately, the distilled product of 
> thinking, maybe license fees won't go away, because it is often easier to 
> pay than to think.
>
> The semantic web is what we have all been groping for since the first 
> computer program was written.  We didn't want text processors; we wanted 
> thought recording and retrieval tools.  We didn't want computer-aided 
> drafting programs; we wanted to create models from our imagination.  But 
> we are still in the firm grip of paradigms meant to encode letters and 
> lines in computer memory.  These paradigms are huge sea anchors holding 
> back progress of the semantic web; and they are made more powerful by the 
> business investment (and inertia) in applications that embody those 
> paradigms.
>
> Semantic web is a tough sell.  The business man, nor the common man in the 
> street, really doesn't care how the application is built, as long as it 
> meets his needs.  Returns on semantic web investments will be slow and 
> diffuse.  The semantic web is like the proverbial elephant--a lot of 
> different things, not the same to everyone.  Part of it is patched on to 
> the old web to provide additional functionality; part of it is webified 
> AI; part of it is just common sense (universal identifiers for resources? 
> what a concept!); part of it is greenfield technology.  So what are you 
> selling, and to whom?
>
> Good luck with your article. I look forward to reading it.
>
> --Paul
>
> Danny Ayers wrote:
>> In brief, some time soon I'm planning to do a moderately comprehensive, 
>> fairly non-technical write-up on this topic ...
>

Received on Tuesday, 1 July 2008 06:15:30 UTC