Re: [Senseclusters-users] using sense clusters to find semantic classes

[Amruta D. P. <am...@cs...>]

> => $ wordvec.pl --wordorder nocare --feats
> 	feats.txt --dims dims.txt  wordpairs.txt > firsttry.vec

Looks good.

>  3. Cluster these word vectors with (give the output of step 2 to)
>  vcluster program in Cluto http://www-users.cs.umn.edu/~karypis/cluto/
>  to get clusters of words !

I assume you have downloaded and unpacked CLUTO. And have set path
to Linux/Solaris directory in CLUTO that contains the programs
vcluster and scluster.

After that, you can simply run

vcluster firsttry.vec #clusters

You can optionally specify the parameters for --clmethod, --crfun,
--rlabelfile options. --rlabelfile will take the feats.txt file
produced by wordvec. You can check Demos/toolkit.sh for help on
how to use the various options in vcluster/scluster.

The above will create the clustering solution file that shows
the cluster id to which each word in feats.txt belongs... Currently
we do not provide any script that will display groups of words
belonging to the same cluster... But, hopefully, thats simple to
write  ... See cgi-bin/format_results.cgi for ideas on
how to achieve that.

Hope this helps somewhat. Please let us know if any questions
are still unresolved/unclear.

Thanks,
Amruta