[Top] [All Lists]

Re: [ontolog-forum] Ontology-based database integration

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Ed Barkmeyer <edbark@xxxxxxxx>
Date: Wed, 07 Oct 2009 13:48:18 -0400
Message-id: <4ACCD462.5080407@xxxxxxxx>
Len Yabloko wrote:
> Thank you for the references. I consider OntoBase an evolution of these 
>existing methods.     (01)

That is essentially my assessment.  My point was that the fact that the 
Semantic Web, as you understand it, doesn't deal with database 
integration says nothing about the novelty of your approach.  In effect, 
you are only arguing, and probably incorrectly, that the buzzword hasn't 
yet been extended to what you are doing.    (02)

> In fact, OntoBase generates semantics models in Protege frame format which 
>can be directly translated to OWL inside Protege.     (03)

Exactly.  This how what you are doing is a step forward from the 1980s 
efforts in the same area.    (04)

> However,  the original link given by Kingsley that I was referring to as 
>"Semantic Web initiatives" was not about using OWL for semantic modeling. Here 
>it is again:
> http://esw.w3.org/topic/Rdb2RdfXG/StateOfTheArt
> Let me quote from the introduction:
> " the role of RDF as an integration platform for data from multiple sources, 
>primarily in form of RDB, is one of the main motivations driving research 
>efforts in mapping RDB to RDF"
>       (05)

And I didn't respond to that, because it is irrelevant, and not only to 
what you are doing.    (06)

> Of course Semantic Web technologies are expanding now into these old 
>territories you have referenced with such detail and accuracy. But one has to 
>be very careful not to confuse semantic modeling with data modeling and data 
>linking with data integration. And ,indeed, not to confuse model-theoretic 
>semantics (developed by Tarski for first order logic and claimed by RDF) with 
>various other semantics. 
>       (07)

All of this is true, but I only understand the relevance of the first 
sentence.  The problem with the rest is that one has to define all those 
terms clearly in order to make any such distinctions, and I'm fairly 
sure that "model-theoretic semantics (a la Tarski)" is the only one on 
which you will get much agreement.  And the link between Tarski 
semantics and "database integration" is surely in the eye of the 
software/knowledge engineer.    (08)

> Yes, and that is one reason not to assume that Semantic Web is fully equipped 
>to deal with data integration. I believe OWL itself is an attempt to go around 
>the inherent limitations of Semantic Web. But simply attaching the word 
>Ontology to Semantic Web does not quite addressed the issue. So I would prefer 
> ontology to be its own independent area of research rather than having 
>secondary role given to it by Semantic Web.
>       (09)

This paragraph means nothing to me, because I don't know what you think 
"Semantic Web" means.  "Knowledge engineering" is a domain of both 
research and practice.  That domain can be divided by distinguishable 
techniques, but you will probably find a few unclassifiable hybrids.  
What is primary or secondary depends on the speaker, his domain of 
expertise (if any), whom he sees as his peers, and his existing or 
would-be funding sources.  If you listen to enough speakers, you will 
realize that "Semantic Web" covers whatever knowledge engineering they 
are doing, and a great many data manipulation things that have nothing 
directly to do with knowledge engineering, such as "triple stores".    (010)

>>> The main goal of data integration should be maintaining consistent mapping 
>between different semantic models. The main challenge is to define what 
>"consistent" means. 
>> Uh, yeah.  The statement of the task strikes me as an oxymoron.  I would 
>> have thought that your OntoBase was about linking multiple data 
>> organizations and representations to a common semantics, rather than the 
>> other way around.  Perhaps I misunderstand...
> The main problem is to arrive to "common semantics". To accomplish that one 
>needs to consider alternative semantic models compatible with data models 
>enforced by data sources. This is what OntoBase allows you to do. But you 
>can't stop even after arriving to mutually compatible model, because new 
>sources can be added and new users may introduce new use cases (or abuse cases 
>,if you insist) 
>       (011)

My concern is that when you have people using the same database field 
for importantly different properties, you can't attach any unifying 
semantic model to the corresponding schema elements.  Or the compatible 
semantic model of the element is so abstract as to lose most of all 
intents and thereby any utility it might have.  If there is a solution 
to this, it becomes data-dependent.  That is, you can use other 
information units attached to the primary keys to classify the "entity" 
and the "state" being represented and thus the semantics of that field 
for that kind of entity in that state.  But when you have two parts of 
the organization with different interpretations of that field for the 
same entity in the same state, you have an unresolvable semantic 
mismatch -- one of the interpreters will be misinformed by the data.  
That is when the "new use" becomes an "ab-use".  And you will find that 
some organizations will knowingly permit such misinformation flows, IF 
they don't perceive it as having significant business impact.  ("Once it 
is in work, it is all right if the sales people misunderstand", until 
the "sales people" become the "customer relationships people" and use 
that data element in answering inquiries "post order" or "post shipment".)
> <snip>
> I have been following these forum long enough to have heard all kinds of 
>claims about what is Semantics and many of these claims were tying it directly 
>to FOL, like no other logic or semantics ever existed.
>       (012)

I will leave the answer to this to John Sowa, and of course, Charles 
Peirce.  ;-)
This exploder is open to all, and it may be difficult to distinguish the 
wheat from the chaff.  For example, the categorization of this email is 
not clear.  ;-)    (013)

-Ed    (014)

Edward J. Barkmeyer                        Email: edbark@xxxxxxxx
National Institute of Standards & Technology
Manufacturing Systems Integration Division
100 Bureau Drive, Stop 8263                Tel: +1 301-975-3528
Gaithersburg, MD 20899-8263                FAX: +1 301-975-4694    (015)

"The opinions expressed above do not reflect consensus of NIST, 
 and have not been reviewed by any Government authority."    (016)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (017)

<Prev in Thread] Current Thread [Next in Thread>