Dear James,
I am sending you a slightly modified version of the search/associate.c
program. This new associate.c differs from the one in the infomap package in
that it also accepts a seed *vector* as a query. In order to query for a
vector you have to call associate with the "-q" (for vector) option instead
of "-w" (for words) or "-d" (for documents).
associate expects the vector to take the form of a simple list of SINGVALS
float numbers, where SINGVALS is the dimension of the reduced word space. It
is defined in admin/default_params.in.
To obtain the words most similar to each dimension of the reduced word
space, you can simply call associate -q with the SINGVALS-dimensional
standard basis vectors.
I might have used an old version of the infomap software when I added the -q
option. You may want to run a diff on the two programs. Let me know if you
need further help.
Best wishes,
Beate
2008/3/7, James Evans <je...@uc...>:
>
> Hi,
>
> I don't know if this is still being maintained, but I and my class (I
> teach sociology at the University of Chicago) have found the software
> useful--I was at Stanford when it was developed and knew Dominic
> Widdows. I was wondering if there is an easy way to print the words
> highest on each of the various dimensions, and the documents highest
> on each dimension. All of the associate queries are documents and
> words relative to each other...and its not obvious to me how to get
> the complete (or even the ranked) word list relative to each dimension.
>
> Thanks,
>
> james
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> infomap-nlp-users mailing list
> inf...@li...
> https://lists.sourceforge.net/lists/listinfo/infomap-nlp-users
>
|
