[Corpora-List] HiDEx version 0.03 released along with a sample vector set and word list.

[Cyrus Shaoul]

Dear Fellow Corpora List members:

A new version of our implementation of the HAL model is now available, 
and for those who do not wish to process a corpus,
  a new sample vector set has been released.

HiDEx is released as GPLv3 source code for Mac OS X and Unix (no 
Microsoft Windows version available). It is available here:

http://www.psych.ualberta.ca/~westburylab/downloads/HiDEx.download.html

(There is a link to the documentation on this web page. Please read it 
before downloading the software.)

The changes since version 0.02 are:

1) Input word lists may now be in uppercase or lowercase.
2) Threshold size calculations are dynamic and are compatible with all 
the available similarity metrics.
3) Other minor bugs fixed.

Our new vector set was made using the Westbury Lab Wikipedia corpus 
(which is also available for download on our site). It has vectors for 
over 50,000 words, and can be used with HiDEx to calculate word-pair 
co-occurrence similarity, word neighborhoods and other metrics. When 
using the sample vectors instead of a corpus there is no way to adjust 
parameters such as window size, window weights or vector normalization 
method.

The vector set is available here.

http://www.psych.ualberta.ca/~westburylab/downloads/HiDEx.vectorset.download.html

Please make sure to download and compile HiDEx before using this vector set.
(The vector set is 1Gb in size, compressed, so please use BitTorrent 
when downloading unless you have access to the Internet2).

Also, we have made a list of over 50,000 English words with their 
neighborhood densities calculated using the Wikipedia corpus. It is 
available at:

     
http://www.psych.ualberta.ca/~westburylab/downloads/westburylab.arcs.ncounts.html

Thanks,

Cyrus


-- 
=[=]={=}=[=]={=}=[=]={=}=[=]={=}=[=]={=}
Cyrus Shaoul
http://www.psych.ualberta.ca/~westburylab/
University of Alberta
=[=]={=}=[=]={=}=[=]={=}=[=]={=}=[=]={=}


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