Hi David,
I am not talking about academic stuff, but about APPLIED ontology.
In my 40 years experience in practical Engineering applications in industry we developed a kind of Formal English that works.
We started from the (rather) unambiguous world of data modeling, and data in databases, not with the ambiguity of natural languages. It is unambiguous enough to build at least parts of refineries conform that data.
We developed a big dictionary-taxonomy-ontology (partly an ISO standard now) that defines a Formal Language (Gellish Formal English, with a variant in Dutch).
The language on our shop floor, being entered as data in databases, uses words for concepts that appeared to have definitions that perfectly fit into a hierarchical taxonomy. That hierarchy enables us to do an automated semantic verification of the expressions.
You may have missed something.
With kind regards,
Andries
Van: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] Namens David Eddy
Verzonden: vrijdag 31 augustus 2012 16:19
Aan: [ontolog-forum]
Onderwerp: Re: [ontolog-forum] Accommodating legacy software
Andries -
On Aug 31, 2012, at 6:18 AM, Andries van Renssen wrote:
To be semantically unambiguous... the language I refer to here is unnatural English. It is the language in & around the software systems that run our world.
I am NOT referring to the English/language that is found in formal dictionaries that have been vetted by professional lexicographers.
If we really want to achieve that systems talk a common language for interoperability and data integration,
Folks have been tilting at that windmill for a looooooong time. Not much to show for it.
A different approach is required.
Everyone—from CEO to clerk—comes to the table with different language. If the CEO attempts to impose their language on the clerk, there may be some temporary success since the clerk wishes to retain their job. An attempt by the clerk to impose their language on the CEO will go no where.
The differences of these languages needs to be embraced, not stamped out.
we should develop and use a kind of Formal English (and other formal natural language variants) in which the semantic rules of the formal language are defined. This requires a managed formal dictionary-taxonomy.
Horse left the barn years ago & is far down the road.
Language as used on the shop floor is NOT a hierarchical taxonomy.
Each & every system—a "typical" organization (I think in terms of Fortune 500)—will have several 1,000s systems, which are the result of the organization writing systems, buying packages & M&A activity. The language in & around this motley collection—very little of it formal—is all over the place.
(Sorry to repeat)... to make this point, I built a tiny prototype "dictionary/thesaurus" (would make the lexicographers cringe) with 2,000 "terms" (in practice the difference between a "word", an acronym & an initialism is ephemeral at best) & 68,000 meanings.
If I were industrious I expect the 34 meanings per term would be in the 68 to 102 range.
REQUIRED reading on this topic is George A. Miller's "Ambiguous Words"
As always, it is left as an exercise for the student to understand who George A. Miller is & what he represents.