Re: Commercial/Real-world Semantic Web Services?

Let us distinquish two different things: semantic technology and semantic web as a platform.

It is obvious semantic technology can have many, many applications.  In my opinion, the major barrier to commercialization is that most object oriented programmers are not well-verse in semantic technology.  To make it work is to have a semantic-object language that combines the execution and reasoning capability.  If it is backward compatible with Java, it will be even better.  Actually I have been working on this on-and-off for 6 years.  If anyone is interested in this endeavor, let me know.

Semantic Web has specific goal to be a platform that enables any two collaborating applications/databases work together regardless of their syntactical interfaces.  The last part is the difference between RDBMS and Semantic Web on the data side, and CORBA, Web Services, and Semantic Web Services on the application side.  I have been working on them for ten years.

Even though we made tremendous progress from ten years ago: RDF, OWL, OWL-S, SPARQL, and many of ontologies, by historical standards, semantic web are still far away from practical applications both on technology and on platform.  There are simply too many missing pieces.  It is simply too costly for a business to fill the gap to produce a product.

Cheers,
William Wu  



----- Original Message ----
From: David Martin <martin@AI.SRI.COM>
To: Bijan Parsia <bparsia@isr.umd.edu>
Cc: Ed Addison <ed@teradisc.org>; Minsu Jang <minsu.jang@gmail.com>; public-sws-ig@w3.org
Sent: Tuesday, October 17, 2006 12:17:49 PM
Subject: Re: Commercial/Real-world Semantic Web Services?


Bijan Parsia wrote:
> 
> On Oct 9, 2006, at 6:13 PM, Ed Addison wrote:
> 
>> I would suggest that those commercial applications that use semantic 
>> web, or semantic web-like technology would not necessarily advertise 
>> that that's what they are doing.  The semantic web is a tool, not a 
>> product or market. SInce the semantic web is in its infancy, 
>> commercial applications that do use semantic web technology most 
>> likely use a significantly scoped down subset of it.  The semantic web 
>> is more likely to slowly infiltrate various information products and 
>> web services rather than suddenly get commercial adoption.  Might be 
>> tough to find or even classify the cases for your study.  Good luck.
> 
> One must be especially careful about such suggestions. While it is true 
> that companies using SWS tech may not have a reason to advertize that 
> (esp. if it is not their product!), it runs a bit close to the "there 
> *is* stuff going on *because* we don't know about it".

Didn't strike me that way.

> There are enough 
> people interested that I would expect *some* information to leak out. In 
> any case, it's best to be humble :)

Well, I mostly agree with the above.  But another form
of humility is to recognize that there may be stuff going on that the
community doesn't hear about, in spite of leakiness.  For example, I am
aware of some serious investigations, uses, and plans to use SWS
approaches within defense and intelligence
agencies.  Many of these folks are not motivated to publish.  Some are
strenuously expected to keep quite, and others will keep quite even if
they don't have to, to avoid any possible hassles.

However, I hasten to add that I'm not aware of any current uses that I
would call "commercial" or "production".  But I am aware of a few
planned uses in the "production" category.

[Gee, I *really* hate to sound like certain politicians who justify 
decisions by reference to secret stuff that's going on.  But anyway, I'm 
not trying to justify anything here, just passing on some information, 
albeit vaguely.]

The other point related to Ed's post is that, in the absence of 
standards or mature tools or a widely adopted methodology it's a lot 
harder for a use of a SWS approach to leak out, simply for the reason 
that it's harder to say what's a SWS approach.

> 
> Also, the original criteria weren't clear. For example, are XACML and 
> WS-Policy "semantic web like" technology?
> 
> But to answer the original poster, I personally don't know of any 
> (successfully) commercial or production uses of OWL-S, WSMO, or the 
> like, at least off hand. I wouldn't take that as conclusive, but I do 
> take it as not a healthy sign.
> 
> I did work with Fujistu on Task Computing (taskcomputing.org). I don't 
> know if that would help.
> 
> At the SWS workshop of the w3c I did an informal poll of the 
> participants asking if their organization was going to spend, oh, 1 
> million dollars in the next year on SWS...I don't think anyone bit (you 
> could look up the minutes). 

Unfortunately that wasn't captured in the minutes or the summary.  I
don't think anyone bit directly, but I think there were some
folks who indicated their organizations might spend that much in an
indirect fashion - e.g., investing in labor for research or to build 
prototype tools and infrastructure.  And I'm pretty sure there were some 
who said their orgs would invest that much if there were standards and 
more-or-less established methodologies for working with them.  (Pretty 
big "if", admittedly.)

In any case, I also recommend to the original poster to take a look at
the papers and presentations from the W3C FSWS workshop, which are 
linked from here:

http://www.w3.org/2005/01/ws-swsf-cfp.html

For example, the INRIA paper (Gandon et al.) describes some interesting 
systems prototyped for business environments.

There are of course a number of relevant workshops to browse, such as these:

http://www.ai.sri.com/SWS2004/
http://www.ai.sri.com/WSS2005/
http://events.deri.at/semantics4ws2006/

... mostly academic but with a smattering of industry work that can help 
to indicate where these technologies might turn out to be useful.

The Int'l. Journal of Web Services Research, which has a strong 
commercial orientation, has so far published about 5 articles that could 
be considered Semantic-Webby.

Cheers,
David


> Some people suggested that they would be 
> involved in a SWS startup in the next 5 years, but I had reason to take 
> that with a grain of salt. Some one at that meeting suggested that 
> people might keep their use secret and thus there was a lot more use 
> than we knew about (which is what makes me hyper trigger about Ed's 
> suggestion).
> 
> Another (perhaps mitigating?) point is that I had a conversation with a 
> microsoft standards weenie wherein I was championing SWSs and his line 
> was that MS was trying to get the lower levels of the stack firmed up 
> (look how long WSDL took!). There was a lot of frustration with the 
> poorness of interopt e.g., with WSDL 1.1 (see the existence of WS-I).
> 
> So, perhaps the time hasn't been ripe. The WS-Policy group wasn't 
> interested, at the time, in the formal semantics I offered or the kinds 
> of expressivity I wanted to add, hence I left the group. However, they 
> had a very narrow goal, to rush a lightly enhanced version of WS-Policy 
> "1.5" into rec. This is both good and bad, yea? It meant that the group 
> cares about more expressive metadata for services and wants it solidly 
> deployable. OTOH, it means innovation is hard to insert. Makes me regret 
> I didn't work with the ad hoc WS-Policy group before the submission to 
> the W3C.
> 
> Cheers,
> Bijan.