Re: Interoperability and Vagueness
Boyan,
The problem of the multiplicity of word senses
has little to do with natural language itself,
and *everything* to do with the number of concepts
(or predicates) needed in the ontology.
> It seems that contextualization, metonymy, multiple
> languages in the world, non-compositionality, other
> difficulties pointed out in this thread, all seem
> to point in the direction of complex and dynamic
> mappings between words as used in a particular
> utterance in a particular language, and ontological
> concepts or an ontologically-anchored representation
> of the meaning of that utterance.
Those problems are certainly harder than the problem
of compiling a programming language. But fairly good
methods are available for all of them. Even the
problem of multiple languages in the world would be
much simpler if there existed a finite ontology from
which all of them selected some subset of concepts
to talk about.
If the ontology were finite, we'd have a fixed upper
bound on the number of word senses. But as Cruse,
Atkins, Kilgarriff, and many others have observed,
there is no upper bound.
The difficulty is caused by reality itself, i.e,
all this physical stuff we move around in and
interact with. That is a problem for ontology
directly and only indirectly for NLP.
Bottom line: NLP with a finite ontology would
be a solved problem by now. Unfortunately,
the real world keeps getting in the way.
John Sowa