[Top] [All Lists]

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

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Ed Barkmeyer <edbark@xxxxxxxx>
Date: Tue, 06 Oct 2009 18:02:47 -0400
Message-id: <4ACBBE87.3020803@xxxxxxxx>

Len Yabloko wrote:
> As I said - OntoBase is designed to simplify data integration and database 
>application development by taking advantage of ontology-based semantic 
>modeling. Part of the method used in OntoBase is semi-automatic generation of 
>ontology guided by the user, who *is* the main source of semantics. Database 
>metadata is used to ensure consistency of the end result and , at the same 
>time, to provide automatic translation of the application activity into 
>database transactions. None of these goals was ever set or accomplished by the 
>above mentioned Semantic Web initiatives. 
>       (01)

They were, however, the objective of several "semantic database 
integration" and "semantics-based federation of distributed databases" 
projects of the 1980s.  As I remember, there was an entire issue of the 
ACM Journal devoted to such projects, long about 1988.  Of course, the 
integrating ontology wasn't called an "ontology" in that time; it was 
called a "semantic data model", with such languages as SDM and OSAM*.  
(Sudha Ram of U. Arizona published a compendium of "semantic data 
modeling concepts" in 1990, with a bibliography of the tower of Babel of 
that time.  I rather shamelessly plagiarized her concepts list for my 
presentation of "OWL as a Data Modeling Language".)  NIST built similar 
tooling back then, and we later used a commercial tool called ORION that 
was a spinoff of some DARPA project (brainchild of Amit Sheth and some 
other famous names).  (More lore from the ancient world before PDF, but 
not nearly so interesting as the Dead Sea Scrolls.)    (02)

The major difference in using OWL as the language for the integrating 
model is that it is a standard language that is well-defined and comes 
with a variety of supporting tools for validating the ontology itself.  
In addition, it offers the possibility of extension to additional 
information bases by linking to additional ontologies, assuming that the 
"integrating ontology" is one, and is not just the OWL version of a 
purpose-built SDM model.  (And it is easy to make tools with much nicer 
GUIs now.)  Unfortunately, many/most databases are treated as having 
inherent "closed world" semantics, which may make it difficult to get a 
"semantic web language" to support some important semantics of the 
databases.    (03)

Ian's point is well-taken:
>> I spent years mapping legacy data into "semantically richer" formats ...,   
>and it always comes down
>> to the data, not the data model. You can guarantee the data modeller got it
>> wrong when they built the system, and the users have been working against
>> rather than with the model for years. Unless you have a good sniff around in
>> the data, you'll never get a real picture of the true semantics (i.e. what
>> the data is actually referring to in the real world). ...
>>         (04)

The data modeler needn't have got it wrong, initially.  The problem is 
that the databases are, for various reasons, rigid, and are not easily 
evolved when the business practices or business environments change.  
So, over time, the schema becomes gradually more and more out of sync 
with the reality of the business, and some bits no longer have any 
vestige of their original intent. (I am reminded of the executive who 
told me: "I know that is what the DB document says, but what my staff 
puts in that field is X, and we are the only ones who still use that 
field."  Unfortunately, he was wrong about the last.)    (05)

Len wrote:
> The project I do for DoD calls this process Semantic Model Discovery. 
>Actually, there are many models some times used in turns, and they all are 
>evolving.     (06)

Yes.  That agrees with my experience.  In a big enough organization, 
there is more than one (mis-)use of certain DB elements, and over time 
some will no longer be used and others will appear.  But the schema will 
live on.    (07)

> They main goal of data integration should be maintaining consistent mapping 
>between different semantic models. The main challenge is to define what 
>"consistent" means.     (08)

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...    (09)

> This where Chris Partridge's work first opened my eyes on true mature of 
>semantics, as opposed to mathematical notion of 1st order logic some times 
>confused with semantics (I know this statement will provoke discussion)
>       (010)

I have seen a lot of things confused with "semantics", because the 
interpretation of "semantics" is usually "what this means for _my_ level 
of abstraction, purpose, and domain of interest", and it often makes the 
assumption that "my" is synonymous with "our common" or "everyone's".  
But anyone who confuses formal logic with semantics understands neither.     (011)

Formal logic is a means of expression, in which the basic elements of 
the language have well-defined meanings and a set of well-defined 
manipulations retain those meanings, and nothing more.  As to whether 
the snark actually is a boojum, classical formal logic offers only two 
possible interpretations:  it is either true or it is false, and it 
cannot be both.  And if you make a set of statements that allow the 
well-defined manipulations to produce a result of "true", then it does.  
That's all there is; there ain't no more.  Formal logic is about "how to 
think"; it is not about what you think about.    (012)

-Ed    (013)

P.S. Ian:
>> I never had a formal method for re-engineering data in those days. In recent
>> years I've been using Chris Partridge's BORO Method. It seems to work pretty
>> well as far as I can tell. Don't tell him I said that though.
>>         (014)

Since Chris is a frequent contributor to this exploder, your secret is 
out...    (015)

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    (016)

"The opinions expressed above do not reflect consensus of NIST, 
 and have not been reviewed by any Government authority."    (017)

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    (018)

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