Date: Thu, 26 Mar 1998 12:15:40 -0500 (EST)
From: "David L. Gants" <dgants@english.uga.edu>
Subject: WordNet and MindNet
>> From: Philip Resnik <resnik@umiacs.umd.edu>
Douglas Beeferman wrote on WN-USERS:
> I think that the sort of data-derived relations computed by MindNet
> are innovative, and that the community can benefit from this approach
> without having to "switch" to any proprietary system. I am unaware of
> MS Research ever having acted to the exclusion of other semantic
> network research, and in fact they have a track record of aggresively
> publishing their results.
Ken Litkowski replied:
> I share Doug's sentiments and revel in the advances that Microsoft has
> made...
> The only alternative is WordNet (and perhaps the Unified Medical
> Language System being developed at the National Library of Medicine).
> George Miller, in a conversation with me a few years ago, said that
> practical considerations, not theoretical, limited the number of
> semantic relations that could be included in WordNet. These
> limitations obviously continue within the limited and declining
> resources available to the yeoman efforts of the WordNet staff.
>
> Given that Microsoft's use of MindNet is clearly ever-expanding and
> probably more and more useful now that they have the critical mass,
> isn't it time that more resources be devoted to enabling the rest of the
> CL world to participate in exploring the ramifications?...
I'm fully in favor of more funding for work on lexical resources, and
I think there are definitely good reasons to look beyond WordNet, not
least of which is the apparent possibility that it may not reach
version 1.7. But I have to say I'm skeptical at the moment about
MindNet as an alternative -- though I'd be very happy to have my
skepticism proven wrong.
For a start, I wonder if Doug or Ken could please provide references
with respect to Microsoft "aggressively publishing" with regard to
MindNet? My attempts to search the web for papers on MindNet,
including several major search engines, the CMP-LG archive, and
research.microsoft.com, turned up no references to MindNet other than
publicity/marketing material ("Bill Gates Builds His Brain Trust", and
the like), and the "NLP publications" page at Microsoft Research
<URL:http://research.microsoft.com/nlp/nlppubs.htm> has no more than
half a dozen papers that might be relevant to MindNet, none of which
was published later than spring 1995 (other than Steve Richardson's
1997 dissertation).
The contrast between this and WordNet is striking: if you want to see
what the benefits of a freely available resource are for the research
community, compare the above with the original "Five Papers on
WordNet" plus the 146 WordNet-related publications (as of January,
1998) listed at <URL:http://www.cis.upenn.edu/~josephr/wn-biblio.html>.
I believe that if Microsoft would like the rest of the CL world to
participate in exploring MindNet, it would do well to make the
resource easily available with a minimum of encumbrances -- and I
think the resulting level of MindNet-based R&D activity would more
than repay Microsoft's investment. (Truth in advertising, though:
doing so would be less in keeping with Microsoft's general philosophy
than, say, with Sun Microsystems' approach. My philosophical bias and
my relationship with the latter would fail a sufficiently sensitive
test for statistical independence. :->)
Meanwhile, let's not forget that EuroWordNet is alive and kicking,
even if the Princeton WordNet has reached the limit of its
resources...
Philip
----------------------------------------------------------------
Philip Resnik, Assistant Professor
Department of Linguistics and Institute for Advanced Computer Studies
1401 Marie Mount Hall UMIACS phone: (301) 405-6760
University of Maryland Linguistics phone: (301) 405-8903
College Park, MD 20742 USA Fax : (301) 405-7104
http://umiacs.umd.edu/~resnik E-mail: resnik@umiacs.umd.edu
-------------------------------------------------------------------------
Humanist Discussion Group
Information at <http://www.kcl.ac.uk/humanities/cch/humanist/>
<http://www.princeton.edu/~mccarty/humanist/>
=========================================================================