Jawit, (01)
I wrote:
>> [I'm sure John did not mean to imply that an ontology would be like an
>> IDEF-1X information model -- having a one-to-one translation to SQL.
>> But many people in this industry think in those terms: that an ontology
>> can be "translated" -- transformed by a rote process -- to an
>> implementation model for 20th century software. I would describe that
>> as wrong thinking. It may be possible for some uses, but it should not
>> be the intent of the ontology. Put another way, that view treats
>> 'ontology' as a synonym for 'information model' or 'object model'.]
>> (02)
Jawit Kien wrote: (03)
> If it is possible, could you clarify what uses that it might be possible to
> use an ontology as an information model or an object model ?
> (04)
Well, suppose our team has developed an OWL model of an application
domain to some level of detail appropriate to the concerns of the
sponsor, and it turns out that every property is class-specific, and
classification is taxonomic -- every class has at most one named
superclass (ignoring all the classifiers whose purpose is solely to
constrain the cardinality of participation in a property). Then that
ontology has a straightforward one-to-one mapping to Java. The Java
model can probably be used directly in a reasoning engine like JESS or
DROOLS. And it can also be used to implement the loading of
corresponding instances (A-boxes) from some other resource(s), and
perhaps in the delivery of the results to humans via a Javascript. (05)
But if the OWL ontology has assignable properties and class definitions
that imply multiple subsumptions, then making a Java model from the
ontology is an engineering effort, and the set of applications in which
that Java model will be used will dictate how the Java model is
constructed and how the implementation elements relate to the OWL
classes and properties. (06)
In a similar way, I can take pretty much any OWL model and create a
corresponding SQL schema that loses nothing but the definitions. But
the blind approach will invent a meaningless primary key (an object-id)
for every A-box/instance. Maybe I will be able to map declared inverse
functional properties to primary keys, but in most cases, some
additional engineering will have to be done to select the primary keys
for the root classes, and the mapping to SQL will greatly depend on
those choices. The question is whether the SQL schema that results from
the rote mapping is really useful for applications. It is likely to be
in 5th normal form -- every property is a separate table, and most
instances are represented by their artificial key in every table. (07)
The more common problem is that we have an SQL schema and a database
full of what will become A-boxes. The problem is to relate the existing
SQL schema elements to the conceptual OWL model of the domain,
particularly when the domain is changing in some way. (08)
-Ed (09)
> Thanks,
> JK
>
>
>> OTOH, an ontology may usefully be "translated" to a different knowledge
>> engineering language, for use by multiple reasoning tools, without
>> significant loss of semantics. That is the distinction I want to make.
>>
>> So, while we are discussing the clarification of vocabulary, can we be
>> more careful about ours?
>>
>> -Ed
>>
>> --
>> 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 Cel: +1 240-672-5800
>>
>> "The opinions expressed above do not reflect consensus of NIST,
>> and have not been reviewed by any Government authority."
>>
>>
>> _________________________________________________________________
>> 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
>>
>>
>> (010)
--
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 Cel: +1 240-672-5800 (011)
"The opinions expressed above do not reflect consensus of NIST,
and have not been reviewed by any Government authority." (012)
_________________________________________________________________
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 (013)
|