[Top] [All Lists]

Re: [ontolog-forum] Terminologies and Ontologies

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Tue, 03 May 2011 12:23:40 -0400
Message-id: <4DC02C0C.5070707@xxxxxxxxxxx>
David, Doug, and Azamat,    (01)

As I said in the note to Azamat,
> My recommendation for anyone who is trying to define anything is
> to check a good dictionary for an independent opinion.    (02)

That doesn't mean you must accept their definition, but it gives
you a starting point that is close to what an educated reader
would expect.    (03)

You may have good reasons to deviate from that definition, but
usually for specializing the dictionary definition to a more
narrow special case.  If you deviate too far, it can become
confusing, and you should probably choose (or coin) another
word or phrase.    (04)

> But what about the very common situation where neither the word/
> phrase nor the definition is in any of the "professional" dictionaries?    (05)

Then you need to write your own definition.  A good methodology is
to adopt a version of what professional lexicographers do:  start
by collecting examples of how the word or phrase is used.    (06)

In the olden days, lexicographers had an army of assistants who would
search printed documents for citations and write them on paper slips.
Today, you can use your favorite search engine to collect thousands
of examples in a few seconds.    (07)

> My prototype dictionary has an average of 34 meanings per word/term.    (08)

That's typical.  You can do exactly what lexicographers did with
paper slips:  analyze the examples and group them according to
the similarity of usage.  The number of groups is the number
of distinct senses you have found.  You might also ask a colleague
to group them and check the agreement or disagreement in groups.    (09)

> One useful way to ensure that those who look at an encoded ontology would
> not confuse the terms in the ontology with NL terms would be to require
> all terms in the ontology to have multi-word names.    (010)

That's a useful practice, since the phrases usually have fewer senses
than single words.  But some words, like 'hydrogen', have fewer senses
then some phrases, such as 'financial institution'.    (011)

> The mapping between NL terms and terms in a single ontology should
> be many-to-many.    (012)

I agree.    (013)

The word 'bank' is everybody's favorite example because it has several
sharply separated meanings derived from different sources.  What's
more interesting is that all the senses seem to be derived from an
old Germanic word, for which the modern English word is 'bench'.    (014)

In any case, lexicographers deal with such examples by creating
several head words, each of which may have several senses.    (015)

For the M-W 9th, there are five head words:  bank1 (river bank),
bank2 (a verb, as in banking a fire), bank3 (financial institution),
bank4 (a verb related to the noun bank3), and bank5 (a specialized
kind of bench or arrangement of benches).    (016)

And by the way, the sense bank3 is a borrowing from the
Italian 'banca', which the Italians borrowed from the
Germanic word for bench, but they specialized it to mean
a table for counting money.    (017)

> Try and choose the best one.    (018)

I agree that a discussion of the full range of variation is useful.
But I would suggest a glossary, in which you state the preferred
definition that you use in the book.    (019)

John    (020)

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    (021)

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