RE: Business Presentation

Ivan-

Thank you for your thoughtful comments. At this point, I really think a community edit is best and therefore have no ownership concerns, thus I take no offense at any criticism.  ;-)

In the spirit of discussion, here are few comments/explanations inline:

Best, -Jeff-

-----Original Message-----
From: Ivan Herman [mailto:ivan@w3.org] 
Sent: Friday, January 18, 2008 5:13 AM
To: Jeff Pollock
Cc: public-sweo-ig@w3.org
Subject: Re: Business Presentation

Hi Jeff,

I have read through the document on the wiki page and, as we agreed, I 
try to start some discussions here...

All in all, I like the approach taken by the document very much. So I 
would say I have some comments and questions within that framework...

- You list Microsoft as a possible partner in 'Partnering choice'. I am 
not sure that is really true. A few weeks ago, when Lee Feigenbaum and I 
were busy collecting SPARQL testimonials, we stumbled across some MS 
usage of Semantic Web[1]. I then contacted them to see if (1) they would 
accept to submit a Semantic Web use case and (2) whether they would 
accept to provide us with a SPARQL testimonial. The response was a clear 
'no' on both accounts. Because this document will be a W3C document, 
eventually, we should be careful of our relationships with members. Ie, 
it is probably wise not to quote them then:-([JTP] 

[JTP] I would not exclude MSFT merely because they couldn't/wouldn't provide an endorsement.  In their Connected Services Framework, they heavily promote their use of RDF for user profile data.  Likewise, recent actions to align with intelidimension's RDF use of sqlserver may indicate some further commitments.  Either way, I do believe that these public commitments to RDF signal that some parts of MSFT agree that RDF is a superior metadata format; mentioning them by name shouldn't cause any disclosure issues.

- I _know_ that was not your intention, but we should be careful about 
the style: reading your piece gives a somewhat negative image of XML... 
(and also RDB...). It may be that my Franco-Hungarian English dialect 
misunderstands you, actually. But, as W3C is also XML (and 2008 will 
include a series of events around the 10 years of XML...), we should 
avoid creating the wrong impression[JTP] 

[JTP] This is one area where I do personally take a firm stance, but am okay with whatever this community decides.  Frankly, I am tired of the XML pundits pretending that it's a data model - the misuse of XSD as a data model is precisely why big corporations must still spend billions on custom development work.  Back in 1999 most people thought that data could easily live outside the relational world, in XML docs, for serious Java applications.  Ultimately, XML is still just slimmed down SGML based on the Infoset 9 level taxonomy - which is fine for document/message markup, but a tragic mistake for "canonical data models."

- W3C does not control SOA specifications. It has done _some_ but, as 
you clearly know, other institutions have done lots of WS-* (and the 
contacts with those were not always, shall we say, 100% peaceful:-). 
Funnily enough, it somehow does not transpire from the text that, well, 
W3C _does_ control the SW specifications!:-)
[JTP] 
[JTP] Great point, we should give OASIS their moment too.  ;-)

- The last section (ENABLING FOUNDATION) makes use of the term 
'metadata' pretty often. I think that W3C has a little bit burned its 
finger with this terms, which contributed to lot of issues around the 
SW. The combination of 'metadata' and RDF/XML gave a fairly one-sided 
image of the SW technologies, and we still bear the consequences. As a 
result, in the last few years we tried not to emphasize the term 
metadata (after all, one person's metadata is another person's data, ie, 
the borderlines are fuzzy...) and put the emphasis more on data 
integration. I wonder whether we could slightly rephrase that section 
along those lines.
[JTP] 
[JTP] I agree that "metadata" is overloaded.  But to the extent that RDF/OWL helps data integration, it's because they are good metadata formats.  Business have plenty of data.  They even have plenty of data integration software.  Depending on whose figures you trust, between $1.5 - $5 billion in license software is spent worldwide on data integration.  And, it's likely that 5x that is spent on professional services for data integration.  But very little of those expenditures can realistically be supplanted by RDF/OWL technology itself. Because the actual software $$$ are going to tools that specialize in certain things like (a) high speed data transformation, (b) federated data queries, and (c) data loading to specific Operational and Analytic Business Applications. All of which has to operate with the terabytes of data that most mid-to-large business already own.  Thus, I would also argue that to simply say that Semantic Web is about "data integration" runs the same overloading risk, and perhaps more so, that the term "metadata" does. 

- This actually touches on a more general issue. I fully understand and 
agree that the paper does not want to go into technical details. 
However, somewhere at the start, it may be worth putting a stake on the 
ground and somehow emphasize that the SW's goal is really on data 
integration. It is there between the lines, your line of arguments uses 
that, but for a slightly outside users some of the statements there may 
not be clear without that. The SWEO group has put some general 
statements on the top of the SW Activity home page[3] (critique 
welcome!:-), and it may be worth taking over something like that at the 
start.
[JTP] 
[JTP] Earlier pontification aside, I do agree that aligning the paper w/other W3C positioning is a good idea.  I will reiterate, though, that positing too narrowly on the "data integration" label may be unwise since there's an implied precision with that term which is both (a) too narrow for the semantic web in the enterprise and (b) too deep for sparql engines and triple stores be a natural alternative


(- By the way, I think Jim Hendler's code was 'a little ontology goes a 
long way', not 'a little RDF goes a long way':-)
[JTP] 
[JTP] Actually, I think I was thinking about "a little semantics goes a long way"  ...I'm pretty sure I've actually heard him use the phrase with both terms,  but now see that the prevailing google citations support the term "semantic."  I even saw one person cite, "the Hendler Principle," nice work Jim!  ;-)

I am sure there will be other comments, but I have to run to a meeting. 
I thought this would be useful in starting up the discussion
[JTP] 
[JTP] good start for a discussion, thank you!  I really appreciate the effort to line things up with other w3c work, I agree that should be an aspect of the version which w3c publishes.  My opinions on XML and Data Integration are obviously a circumstance of having a data integration vendor perspective.

  I do stand by the ideas that XMLs success is tempered by its mis-applications (leading in no small part, to the need for rdf & owl) and data integration is only a small part of what rdf/owl can accomplish for a business (albeit only with much effort right now)... I am curious to hear more of your thoughts.


Ivan



[1] 
http://blogs.msdn.com/imm/archive/2007/11/21/creating-and-extending-owl-ontologies-in-imm.aspx
[2] http://esw.w3.org/topic/SemanticWebTools
[3] http://www.w3.org/2001/sw/

Susie M Stephens wrote:
> 
> Jeff Pollock is taking the lead on creating a business presentation on the
> Semantic Web. He is starting this process by writing a document that
> collects his thoughts. Please could you take a look at the document, and
> provide any feedback that you may have by January 22. The document is
> posted to the SWEO Wiki [1], so you can directly make edits.
> 
> Thanks,
> 
> Susie
> 
> [1] http://esw.w3.org/topic/SweoIG/TaskForces/BusinessPresentation
> 

-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf

Received on Sunday, 20 January 2008 07:08:17 UTC