Message 12254 of the SUO list

Subject: Re: SUO: Multi-Source Ontology (MSO) Draft Ballot Question
Date: Fri, 30 Jan 2004 17:13:55 +1000
From: Philippe Martin
In-reply-to: msg12252 by Richard Cooper
Follow-up: msg12255 by Richard Cooper 


> Philippe, I notice that your paper at the site you reference
> below is titled "Correction and Extension of WordNet 1.7".

To be more precise it should have been titled "Correction and
extension of the noun-related part of WordNet 1.7 for
knowledge representation/retrieval/sharing purposes".

I do think that a more KR oriented approach/methodology (i.e.
simply being slightly more precise) would be beneficial
to everyone (KR people and linguists) but not all members of
the WordNet team are willing to go that way.


> Did your corrections and extensions make it into the latest
> version - WordNet 2.0, or are they still being considered
> by the Princeton crew?

Neither, I guess. Here is the story so far.

On the 25/06/2002, I sent the URL of my lexical/semantic corrections
of WordNet 1.7 (786 corrections at that time) to wordnet@princeton.edu
(i.e. actually Randee Tengi) and late October I enquired about their
insertion in WordNet. Here are 2 extracts from her answers on the
28/10/2002 and 29/10/2002.

  I can't remember exactly what you've sent in the past, but I'm
  sure they were forwarded to the lexicographers for inclusion
  in a future release (please don't resend them).

  We're really just 4 part-time workers here, and we can't easily
  keep up with all the wonderful offers of input, data, corrections,
  etc., that our thousands of users offer us.  We're thrilled that
  they find WordNet useful and applicable, but we don't have the
  resources to get personally involved in others' reasearch unless
  there is a very direct link to the basic WordNet that we provide,
  or they're giving us funding to do something!

WordNet 2.0 (the version including correspondences between 
noun synsets and verb synsets) came directly after the 1.7 and in
August 2003, i.e. much earlier than announced. 
May be the corrections I sent were too numerous or too late to
be included in the 2.0 version but it seems that they have been
ignored (including the reports of link redundancies) although I
have not done a systematic check.

When I presented my article in August 2003 I was advised to contact
Christiane Fellbaum directly. However, I decided to wait a bit
to see if I could integrate WordNet 2.0 (noun-related part and
others) and thus send up-to-date corrections. But this would have
required at the very least a full month of work on this task and
this did not happen. Since I now know that this is not likely to 
happen soon, and thanks to your reminder, I am sending an email
to Christiane Fellbaum right now.


 
> What is the most correct and extended WordNet now available,
> or is yours the best at this point?

For manual knowledge representation and sharing purposes (where
you should not need the verb/adverb/adjective related parts),
mine is more correct and extended, and better to use.

For automatic NLP and other linguistic applications, I think
the advantage of having the correspondences between noun synsets
and verb synsets far outweights my few corrections and extensions
(and a simple approach to generate category identifiers can be
applied).


Philippe