W3C home > Mailing lists > Public > public-swbp-wg@w3.org > November 2005

Re: [SE] Ontology Driven Architecture Note

From: Phil Tetlow <philip.tetlow@uk.ibm.com>
Date: Wed, 16 Nov 2005 13:01:42 +0000
To: Jim Hendler <hendler@cs.umd.edu>
Cc: Tim Berners-Lee <timbl@w3.org>, public-swbp-wg@w3.org
Message-ID: <OF4069578A.1F292E9E-ON802570BB.004036DB-802570BB.0047A458@uk.ibm.com>


You are indeed wise! Nevertheless I must confess to be personally troubled
at this point. Like you I share and am passionate about the Semantic Web
vision. Furthermore, if I were to aspire to any particular label, I would
hope that history would mark me down as a 'proponent' of the Semantic Web
movement. Even so, having spent nearly two decades as a practitioner I feel
that I may have become tainted somewhat. For me there is unfortunately a
significant gulf between the ultimate end game of the Semantic Web and the
historical backdrop from which it’s founding concepts came (as I believe
you also agree). This is nothing to do with validity, theory or the
technology behind the Sem Web, but more to do with the adoption paths that
its ancestral technology movements have historically followed.

Take, for example, AI (as you suggest) or the formal methods community that
blossomed in the mid to late 80’s. Both were beautiful paradigms in
principle and indeed when applied properly can lead to some great and
worthy implementations. Even so neither has experienced significant
take-up, unlike the mainstream Web. So I think the really relevant question
for us all is ‘why’, especially given that the Semantic Web left its youth
behind some time ago?

I would never pretend to be a serious technology or social commentator, but
I think it may be important to be frank about the original Web’s
architecture. As I believe Tim himself has professed many times, the
strength of technologies like HTML is that they can be bent into some
pretty ugly shapes and yet the Web as a whole still bounces back into form.
The point surely about the Semantic Web is that it was designed
specifically NOT to bend? This unfortunately means that a significant
number of anti-patterns that were available for the original Web’s adoption
are not applicable this time around. Hence if we are still aspirant of mass
adoption for the Semantic Web, my personal view is that we will need to
engage in some pretty lateral thinking (both pure and dirty) to fill the
vacuum of opportunity we are pragmatically facing.

Does this mean that we will have to abuse parts of the Semantic Web for the
sake of its survival? Does this mean that cracks will appear in the joining
of the Web with the Semantic Web as you suggest? Who am I to comment,
surely all future realities reside under the control a global audience’s
consensus. Even so I do think that we have to at least consider some ground
that is currently way of the W3C’s radar. I sincerely hope, therefore, that
you will not consider me to be a heretic just because I am in favour of
fighting far off battles in order to win the war? I will happily leave
border protection in the hands of far more capable experts like your good
self ;0)

Best Regards,

Philip Tetlow
Senior Consultant (Certified Technical Architect)
IBM Business Consulting Services

Mail: IBM United Kingdom Limited, 1175 Century Way, Thorpe Park, Colton,
Leeds, LS15 8ZB
Current Assignment: DWP BPRP (Metadata)
Mobile: +44 (0)7740 923328
Email: philip.tetlow@uk.ibm.com

             Jim Hendler                                                   
             du>                                                        To 
                                       Phil Tetlow/UK/IBM@IBMGB            
             15/11/2005 19:20                                           cc 
                                       Re: [SE] Ontology Driven            
                                       Architecture Note                   

>I understand your point and think we may well have a compromise - Ill ask
>Jeff Pan to update the document to read "Semantic (Intra) Web's", if
>OK with you?


>You raise another interesting and important point though. I personally
>of a good handful for applications for RDF, OWL and even SWRL that do not
>directly relate to any 'Web', or intranet,  context, so should such
>practices be discouraged from a standards perspective, or even for the
>of preserving an absolute reference description of the Semantic Web? I
>guess the only real answer is "no" given that the same could be said of
>older Web technologies like HTML. Nevertheless I do think it is important
>to point out somewhere that Sem Web technologies can be extremely powerful
>if applied properly, and equally so in cases of misapplication.

I don't know what you mean be RDF, OWL, SWRL that
don't relate to the Web ... in the sense that if
they're using that language, things have URIs -
if they choose not to make them available via
HTTP that is their loss, but at least in theory
they will be able to link in when they are ready

>As for your point on Sem Web use cases not incorporating RDF or OWL etc, I
>think I agree but am not quite sure yetŠ.In the Software Engineering
>Workshop at ISWC, for example, we had a guy point out that the latest
>version of Java is capable of consuming/using 'metaobjects' (forgive me
>being vague as I have not looked up the specifics yet and am just
>to the raw concept here) and I guess that as long as such objects used URI
>conventions based around some triple-based graph scheme, regardless of the
>nomenclature in which they were instantiated, I can already see a
>for the creation of some powerful 'interfaces' to the Sem Web world
>the use of any of its base language set - I guess my point is that if the
>architecture of a use case is correct, the implementation technologies
>behind it become less important (but I have never been too sure about such
>philosophical matters! ;0)). I know that this is not the same as saying
>that something is applicable to the Semantic Web even if it does not
>contain any trace of its heritage (for example I saw a presentation at
>on logic compression that I did not consider to be overly relevant), but I
>do think that there are some issues here that we may have to think about.

this year I was please that there were some talks
at ISWC that DID actually have Semantic Web (as
opposed to traditional AI) technologies - some of
this has to do with how the EU is funding this
stuff, but at least in theory the SW is defined
by architecture, much the way the Web is.  Part
of why I came down so hard on you is exactly
because I'm fighting this battle to keep the SW
related to the Web and not to become a new name
for the same old KR --because we lose if that

So there's lots of work that could "inform" the
SW, and I take a charitable view that that is why
a lot of this stuff is called Sem Web, but I
think the URI basis of RDF/OWL, and the linking
inherent in its use, are key ideas we need to
focus on.  (btw, the theme for next years ISWC is
putting the Web back in Semantic Web, so this
"meme" is catching on...

>Again your thoughts would indeed be valued.
>Best Regards,
>Philip Tetlow
>Senior Consultant (Certified Technical Architect)
>IBM Business Consulting Services
>Mail: IBM United Kingdom Limited, 1175 Century Way, Thorpe Park, Colton,
>Leeds, LS15 8ZB
>Current Assignment: DWP BPRP (Metadata)
>Mobile: +44 (0)7740 923328
>Email: philip.tetlow@uk.ibm.com
>              Jim Hendler
>              <hendler@cs.umd.e
>              du>
>                                        Phil Tetlow/UK/IBM@IBMGB
>              15/11/2005 02:37
>                                        Tim Berners-Lee <timbl@w3.org>,
>                                        "best-practice list"
>                                        <public-swbp-wg@w3.org>,
>                                        danbri@s3.org
>                                        Re: [SE] Ontology Driven
>                                        Architecture Note
>Phil - it wasn't the concept that bothered me -
>it was the terminology - the idea of Semantic
>Intra-Webs doesn't bother me, the idea of
>"Semantic Webs" (with an S) is what I think we
>need to avoid because it gives people the wrong
>idea - for example, I saw a couple of things at
>ISWC that claimed to be Semantic Web, but didn't
>use RDF, OWL, any kind of URI scheme, or anything
>that would allow linking into the Semantic Web...
>doesn't seem to me that that is a good thing to
>   -JH
>p.s. FWIW, I thought this note would be a good
>one for SWBP, in fact better than many of the
>things turned out that are sort of "ontologies
>for experts" instead of "why use the Semantic
>Web" -- if the SW CG does start a Sem Web
>Outreach WG, then I hope this document will go
>At 21:00 +0000 11/14/05, Phil Tetlow wrote:
>>I fully appreciate your point and indeed agree in part, but I don't
>>necessarily believe that the note is as dangerous, divisive or naïve as
>>indicate (but I am hoping you will correct me further). As the preceding
>>paragraph to the section in question states "it is important to be
>>pragmatic" and also mentions that the two cases presented relate to a
>>"relational standpoint".
>>This section was, in fact, included following some considered and valued
>>input from Alan Rector and I personally think that it adds great value.
>>me, it helps break down a number of barriers for many "traditional"
>>practitioners who IMHO have a somewhat esoteric view of the Semantic Web
>>a holistic reality.
>>The main purpose of the note is to encourage take up in new communities,
>>"bridge building" if you will. As such, even though, we all know this
>>of the bridge that linking "islands" of the Semantic Web will be
>>in the future, I think there is another, equally important point we
>>not forget - if the islands are not constructed in the first place, the
>>idea of linking them together into higher orders of ontology is merely
>>fanciful. Just because such islands choose to exist in the same
>>ocean of HTTP URI space does not preclude them from being located over
>>visible horizon, nor does it suggest that they have to be quarantined
>>the rest of the world to prevent infection (please forgive my somewhat
>>forceful analogy here). Furthermore, I think it is important to note that
>>less experienced practitioners may well be more comfortable with island
>>building, than continent construction right now.
>>There is also a question of purpose here. This note was deliberately
>>written within the Best Practices Working Group to undergo hard review
>>it has indeed undergone a baptism of fire so far), so what exactly do we
>>mean by "Best Practice" here? Is it about the betterment of the notion of
>>"a Semantic Web" as you appear to suggest, or is it more to do with the
>>application of current technologies created under the Semantic Web
>>If the latter is more relevant, then as an industrial practitioner using
>>such technologies in the field right now, the significant feedback I am
>>getting from "corporate" adopters is that they have purely selfish
>>for adoption, wanting to address specific problems in their own specific
>>domains. Hence do we want to refer to such practices using "intra-based"
>>terminologies? My feeling is not for exactly the reasons you mention.
>>users deliberately adopt global namespaces from which to build their own
>>isolated understandings for good reason, and so these understandings may
>>well be blended into the wider community of the Semantic Web at some
>>in the future (as indicated in the second bullet in the list). Therefore,
>>at present, they are neither totally isolated nor totally integrated;
>>are merely invisible web fragments of an early and sporadic semantic
>>Is this bad practice? Is this misconceived? I think not. It may not be
>>pure, but it is practical and it is helping to solve a number of serious
>>and significant real world problems right now; hence the need for best
>>practice guidance in this area. To marginalise such types of application
>>would, in my "naïve" mind, be counterproductive, but I wholeheartedly
>>with your opinion that it is important to send out the most appropriate
>>guidance to an eagerly awaiting audience at this time. Any further advice
>>you might have would indeed be appreciated, in what is undoutably a
>>contentious area.
>>I think that it is also important to reiterate that many of the
>>applications of Sem Web technologies we are now seeing are less concerned
>>with AI and more to do with formality of description and transformation
>>Ranting over. I must apologise!  ;0)
>>Best Regards,
>>Philip Tetlow
>>Senior Consultant (Certified Technical Architect)
>>IBM Business Consulting Services
>>Mail: IBM United Kingdom Limited, 1175 Century Way, Thorpe Park, Colton,
>>Leeds, LS15 8ZB
>>Current Assignment: DWP BPRP (Metadata)
>>Mobile: +44 (0)7740 923328
>>Email: philip.tetlow@uk.ibm.com
>>               Jim Hendler
>>               <hendler@cs.umd.e
>>               du>
>>                                         Phil Tetlow/UK/IBM@IBMGB,
>>               14/11/2005 16:44          danbri@w3.org
>>                                         "best-practice list"
>>                                         <public-swbp-wg@w3.org>, Tim
>>                                         Berners-Lee <timbl@w3.org>
>>                                         Re: [SE] Ontology Driven
>>                                         Architecture Note
>>FWIW, I was planning to raise some objections when this went out - I
>>can do it now -- basically, I think we should remove the discussion
>>of "a collection of semantic webs" which is both naive and misleading
>>(section 3.4 of [1]) -- rather, if you wish to refer to something
>>like "Semantic intra-nets" or such I could live with that -- the
>>point is this content all lives in the same exact address space (the
>>http URI space) and separate documents within corporations or the
>>like, may be protected by firewalls, or by lack of linking, but since
>>they still participate in this same universal space (and via same
>>protocols, standards, etc;) saying "Semantic Webs" is as wrong as
>>referring to separate "Webs" -- the WWW has intranet/intraweb
>>components which are walled-off from others, and this was crucial to
>>early Web development, but it is exactly that these could eventually
>>be linked to others that we have a (singular) World Wide Web, and
>>conveying the idea that somehow the Sem Web is different is both
>>misleading and wrong -- if someone totally foolish wanted to create
>>their own, unregistered URI scheme, keep their ontologies against
>>that scheme (and I guess copy the owl namespace into that space or
>>else they link via owl: concepts), and make sure nothing every
>>touched the rest of the Web it could be a separate Semantic Web, but
>>it seems like an odd and vicious idea to do so.   Linking "islands"
>>of the Semantic Web will eventually be very important to its success,
>>and it is VERY important that we don't convey the idea that these
>>islands are somehow separate -- if we do, then much of the Sem Web
>>technology "degrades" back to the traditional, unlinkable, AI stuff,
>>which is what we are trying to avoid.
>>    Tim BL and I had a fight with one of the EU funders who kept trying
>>to refer to multiple Semantic Webs, and seeing SWBP feed into this
>>foolish misconception would not be a good thing
>>    -Jim Hendler
>>     AC Rep
>>    MIND Lab
>>[1] http://www.w3.org/2001/sw/BestPractices/SE/ODA/

>>At 8:47 +0000 11/14/05, Phil Tetlow wrote:
>>>How do you suggest we go about SWIG review of the ODA note?
>>>It can be found at http://www.w3.org/2001/sw/BestPractices/SE/ODA/

>>>Best Regards,
>>>Philip Tetlow
>>>Senior Consultant (Certified Technical Architect)
>>>IBM Business Consulting Services
>>>Mail: IBM United Kingdom Limited, 1175 Century Way, Thorpe Park, Colton,
>>>Leeds, LS15 8ZB
>>>Current Assignment: DWP BPRP (Metadata)
>>>Mobile: +44 (0)7740 923328
>>>Email: philip.tetlow@uk.ibm.com
>>Professor James Hendler                                  Director
>>Joint Institute for Knowledge Discovery
>>UMIACS, Univ of Maryland
>>College Park, MD 20742

>>(New course: http://www.cs.umd.edu/~hendler/CMSC498w/)
>Professor James Hendler                                  Director
>Joint Institute for Knowledge Discovery
>UMIACS, Univ of Maryland
>College Park, MD 20742

>(New course: http://www.cs.umd.edu/~hendler/CMSC498w/)

Professor James Hendler                                  Director
Joint Institute for Knowledge Discovery                        301-405-2696
UMIACS, Univ of Maryland                                       301-314-9734
College Park, MD 20742

(New course: http://www.cs.umd.edu/~hendler/CMSC498w/)
Received on Wednesday, 16 November 2005 13:02:44 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:09:45 UTC