[Corpora-List] TextGraphs-6 - Call for Papers - ACL-HLT 2011 Workshop

[Thomas L. Packer]

Very good.  

What I wish someone would do is to construct a complete "Graph of Research"
and provide me a tool to mine my dissertation topic and methods from that
graph.  :-)  Call me lazy.

I guess for that to work, it would actually have to infer new (fruitful)
connections in the graph the way some researchers are predicting new
connections in social graphs.  

You'll probably tell me that that's being done already, too.  :-)  I can't
wait to see it.

Thomas L. Packer
~~~~~~~~~~~~~~~~~~~~


-----Original Message-----
From: Rada Mihalcea [mailto:rada at cs.unt.edu] 
Sent: Monday, February 07, 2011 11:09 AM
To: Thomas L. Packer
Cc: 'John F. Sowa'; corpora at uib.no
Subject: Re: [Corpora-List] TextGraphs-6 - Call for Papers - ACL-HLT 2011
Workshop


Oh, already answered :)

Indeed, there is work on eigenvalues on semantic networks -- see all the
research on PageRank on WordNet or other networks.

But I agree there is often a disconnect between the fields (which
sometimes just makes things so much more interesting -- e.g., isn't it
exciting to discover that an algorithm created for managing traffic on
road networks can be so nicely applied to texts?)

Rada

On Mon, 7 Feb 2011, Thomas L. Packer wrote:

>I also wonder why there seems to be a persistent disconnect among
>researchers, whether it be between late vs. early research or even between
>two contemporary disciplines or fields.  More connections should be made
>(although, the more research is being done, the harder it is to keep it
>connected).
>
>However, in this case, a new term makes sense because it refers to
something
>different.  I don't think "graph based methods in NLP" is identical to
>"semantic networks".  Most likely the TextGraphs people would view a
>semantic network as a special case of graph based methods.  Other cases are
>possible, such as the interesting idea of applying spectral graph theory to
>documents or other linguistic constructions where each instance is
>positioned in a metric space and the nearest neighbors are represented as a
>graph.  (For example, Google' PageRank.)  In other words, there are more
>ways to apply graphs to NLP than by using semantic networks.
>
>... but it makes me wonder:  has anyone found use in computing the
>eigenvalues of a semantic network?  :-)
>
>Thomas L. Packer
>~~~~~~~~~~~~~~~~~~~~
>
>-----Original Message-----
>From: corpora-bounces at uib.no [mailto:corpora-bounces at uib.no] On Behalf Of
>John F. Sowa
>Sent: Monday, February 07, 2011 5:52 AM
>To: corpora at uib.no
>Subject: Re: [Corpora-List] TextGraphs-6 - Call for Papers - ACL-HLT 2011
>Workshop
>
>On 2/6/2011 3:10 PM, TextGraphs-6 wrote:
>> TextGraphs-6: Graph-based Methods for Natural Language Processing
>
>The term 'semantic network' was coined at the Cambridge Language
>Research Unit (CLRU) a half century ago for "Graph-based Methods
>for Natural Language Processing".  See, for example,
>
>    http://www.jfsowa.com/pubs/semnet.htm
>
>Why would anybody coin a new term?
>
>There is a tradition in AI and NLP to change terminology every
>10 or 20 years so that all the old R & D can be ignored.  That
>practice establishes a new COTD (Change of Terminology Date),
>which makes it easier to publish and patent "new" ideas.
>
>John Sowa
>
>
>_______________________________________________
>Corpora mailing list
>Corpora at uib.no
>http://mailman.uib.no/listinfo/corpora
>
>
>_______________________________________________
>Corpora mailing list
>Corpora at uib.no
>http://mailman.uib.no/listinfo/corpora
>



_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora