| Computers Forum Index » Computer Artificial Intelligence - Language » More than a definition of a contextual distance among... |
|
Page 1 of 1 |
|
| Author |
Message |
| Albretch Mueller... |
 Posted: Thu Sep 03, 2009 7:03 pm |
|
|
|
Guest
|
More than trying to study the usage patterns of an author or how the
meaning of words change/evolve based on its distribution and usage
contexts, say you want to study how a certain concept evolves/changes -
along with its ramifications-. You don't even have to use certain
words in order to directly or indirectly evoke a concept
How can you do that using corpora?
I think there is a term to describe what I mean. Something like
"concept spaces", what was it?
Any -good- studies on "concept spaces" (and please, I do know I can
use google (or was it google can use me)  )
Thank you very much
lbrtchx |
|
|
| Back to top |
|
|
|
| Don Phillipson... |
| Posted: Fri Sep 04, 2009 12:11 am |
|
|
|
Guest
|
"Albretch Mueller" <lbrtchx at (no spam) gmail.com> wrote in message
news:4dc90ed5-e9a7-4c1d-94e2-fbffb52cf622 at (no spam) o41g2000yqb.googlegroups.com...
Quote: More than trying to study the usage patterns of an author or how the
meaning of words change/evolve based on its distribution and usage
contexts, say you want to study how a certain concept evolves/changes -
along with its ramifications-. You don't even have to use certain
words in order to directly or indirectly evoke a concept
How can you do that using corpora?
I think there is a term to describe what I mean. Something like
"concept spaces", what was it?
We still do not understand (at AUE) what you seek or want.
We do not recognize "contextual distances" as distinct from
different contexts.
--
Don Phillipson
Carlsbad Springs
(Ottawa, Canada) |
|
|
| Back to top |
|
|
|
| Albretch Mueller... |
| Posted: Mon Sep 07, 2009 12:32 pm |
|
|
|
Guest
|
On Sep 4, 4:21 pm, Ian Parker <ianpark... at (no spam) gmail.com> wrote:
Quote: On 3 Sep, 20:03, Albretch Mueller <lbrt... at (no spam) gmail.com> wrote:
so that the context is insensitive to similes
~
as you pointed out taking similes into account is very important. In
fact this is exactly part of what you need; tracking the usage changes
of different terms
~
Quote: However if you use LSA at very short ranges, what you will increasingly find is style.
~
Any thorough Mathematical introduction to LSA in which these aspects
are discussed? How does the computational complexity changes with
short ranges? Is it an NP problem?
~
Quote: Have I been of any help, or have I given you something to think about?
~
yes you have, yet I still need to find out which other methods do
researchers use when they need the shortest ranges
~
Thank you
lbrtchx |
|
|
| Back to top |
|
|
|
|
