A 'synset' is just a mechanism for representing what the author hopes is a
relatively well-defined concept. Any ontology that has pointers to words
in a human language will form the basis for a 'language that uses synsets',
if the ontology allows pointers to more than one word for each represented
WordNet is used that way by NL researchers, but its shortcomings in that
role have been much remarked. It was originally intended to represent
cognitive insights into the use of language, but was pressed into service
for NL research because it was free and was fairly comprehensive for
English, and had a hierarchical structure unavailable in any other free
resource at the time it was first used. I think that NL research needs a
reworked 'WordNet' with the concepts better structured for automated
inferencing. That is one of the functions that I hope the COSMO (or some
successor) will serve, so I am adding in pointers to WordNet synsets as well
as the LDOCE defining vocabulary, but there is no simple mapping of a good
ontology to WordNet because the WordNet structure is not based on principles
of inheritance; so a simple 'mapping' of WordNet to an ontology like Cyc or
SUMO is of limited usefulness, and does not correct the problems. (01)
Efforts to rework the WordNet have been made by groups such as Ed Hovy's and
Martha Palmer's and Nicola Guarino's, but they haven't yet been extended to
create the ontologically correct variant of the WordNet. That's a lot of
work. What I am trying to do is to create a version of COSMO that at least
maps the basic words representing the semantic primitives (and some other
useful basic concepts), using the LDOCE defining vocabulary as a plausible
starting list that includes most of the semantic primitives. This will
allow experiments with NL, confined to the basic vocabulary, but aimed at a
deeper level of semantic representation than is possible with the WordNet.
This is one tactic I want to use to approach the goal of getting a computer
to have a meaningful conversation with a 6-year old. (02)
Among the problems with the current WordNet, from an ontological
(1) in linguistic use a single word is often used to represent some
general concept and also a more specific concept. WordNet does not allow a
word to be a hypernyms of itself, so they cannot represent such relations.
(2) different synsets containing the same word are often indistinguishable
in logical meaning. The NL people refer to this by stating that WordNet is
'too fine-grained' - a polite way of saying they can't figure out the
differences in meaning between some different synsets (and I can't either,
so I conclude that they are in fact logically identical or largely
overlapping in meaning).
(3) some synsets include more than one meaning that are clearly
distinguishable logically. This means that just aggregating WordNet senses
will not create the hierarchy that is optimal for NLU.
(4) the meanings, or parts of the meanings, of synsets that have no word in
common can also overlap or be identical.
(5) the representation of an Event in COSMO will often map to both a noun
and a verb (e.g. move and motion). The linguistic/syntactic distinction is
not necessary at the logical level, and is best left as a task for the NLU
program to handle rather than the knowledge representation. In WordNet they
are different synsets.
(6) the hierarchy is, in many places, not properly structured for
inheritance of properties, and cannot be used for accurate logical
But WordNet still represents a tremendous and useful effort, and is useful
for NL at a shallow semantic level. It is a good start, but something
similar with a more precise semantics is needed. (04)
> -----Original Message-----
> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-
> bounces@xxxxxxxxxxxxxxxx] On Behalf Of Richard H. McCullough
> Sent: Monday, April 20, 2009 4:11 PM
> To: [ontolog-forum]
> Subject: Re: [ontolog-forum] web-syllogism-and-worldview
> How about a language that uses synsets instead of words?
> Do you know if anyone has researched that?
> A synset is an equivalence class, similar to your
> definition of proposition.
> Dick McCullough
> ----- Original Message -----
> From: "John F. Sowa" <sowa@xxxxxxxxxxx>
> > The simple reason is that mathematics of any kind (including symbolic
> > logic) makes statements with sharp, absolute criteria of precision.
> > (Even fuzzy logic and probability theory make precise statements
> > about fuzziness and probabilities.) In the initial unsettled stages
> > of research, such precision is impossible. It's also impossible to
> > quantify the exact amount of fuzziness. That's why ordinary language
> > is far better suited to the *development* of a theory than to the
> > final statement of the theory.
> 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
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
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 (08)