[Top] [All Lists]

Re: [ontolog-forum] Using controlled natural languages for ontology

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Fri, 11 Mar 2011 12:08:47 -0500
Message-id: <4D7A571F.404@xxxxxxxxxxx>
Ed and Mike,    (01)

I agree with both of your notes in this thread.    (02)

> I have described the process of learning "enough about the domain
> to put forward a model which is largely complete" as 'osmosis'.
> The knowledge engineer talks to enough domain experts and asks questions
> questions and gradually forms an image of the basic model.    (03)

Yes.  And I would point out that the expert KEs who have invested
that much time and effort have reached the apprentice level in
the subject domain.  They can't replace the SMEs, but they can
serve as translators between the SMEs and the computer systems.    (04)

>> And of course having a
>> model format which does not require them to learn some language.    (05)

> Ultimately, they do have to learn the language of the model.  If nothing
> else, there will always be terminology issues.  The value of a CNL is
> that it can make that learning process much easier.  It is a small step
> from a language they speak.    (06)

I completely agree.    (07)

> Over the years, I have found that domain experts can also easily
> understand simple diagrams in UML, NIAM/ORM and BPMN, and the first two
> can be readily transformed into formal ontologies.  The trick is not to
> introduce more than 5 concept-symbol pairs and talk about them at the
> thing/relationship/action level.  But CNLs allow for presenting more
> elaborate rules, both 'structural' and 'operational' (in Ron Ross's
> terminology).    (08)

I have always been a big fan of graphic notations.  And I would
add that SMEs often have elaborate diagramming conventions of their
own, which the KEs should learn and use.    (09)

One more point:  The CGIF notation for Common Logic is designed to
represent unordered collections of nodes and arcs -- with the option
of grouping such collections in nested packets (called contexts).    (010)

That representation is sufficiently general to serialize a large
number of common diagrams -- ranging from electronic circuits
to process flow diagrams in chemical engineering.  It is then
possible to write rules in CGIF or CLIF for mapping those diagrams
to other versions of logic and to CNLs.    (011)

John    (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)

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