Matthew West wrote:
>[MW] Dear PatC,
>So much harder to cope with?
Generalizing, are you
postulating use of a few words (by, of, to, on, …), perhaps
exactly the set of
prepositions, as an NLP representation of binary relationships
Aren’t there cases
where the preposition itself has multiple meanings? For “of”:
The book of Mark
The Duke of Windsor
Sort of like a bird
Lots of biscuits
A cause of global warming
These examples seem to
display a diversity of meanings. Why would your
approach make it easy to
handle these relationships? Similar concerns
about the other
prepositions would also surface IMHO after analysis.
[MW] I was just being lazy. If I were working this out
properly (as we have in ISO 15926) I would have a set of role terms that were
properly defined. I was only trying to make a point about understandability of
the binary form, which need be no harder than an n-ary form if you present it
The “understandability” issue
is key here. In a commercial project, the entire team must have nearly
identical understanding of the meaning at issue. If one uses a
preposition “of” as a link in a database relation, all members of
the team must have the same meaning in mind to be able to use it. The
team won’t have time to philosophize on the words, and they will almost
certainly want to use a previously existing ontology, even if it is small.
So using polysemous prepositions won’t work well in that ecosystem.
Instead, a single sense meaning must be defined for each link word.
That’s why I don’t think using
binary forms has the same expressive power as the signature, such as
give( Donor, Thing, Recipient) where give
is essentially monosemous (Sp?).
Rich AT EnglishLogicKernel DOT com