Table of Contents
ssutil -- tool for inspecting semantic spaces
ssutil command [...]
ssutil help sub-cmd
ssutil fileinfo file
ssutil show sub-cmd file
ssutil compact infile outfile [flags]
ssutil search file k term
ssutil list file doc
ssutil coords file [flags] [term | doc]
ssutil odist file [opts] [term | doc]
ssutil displays various information about a semantic space and offers
rudimentary support for searching.
The commands and their respective sub-commands/options are as follows:
- help cmd
- display help for a given command cmd.
fileinfo file
display info about a particular space file.
show sub_cmd display indexes or number of dimensions. Valid values for
sub_cmd are:
- k
- show the maximum number of dimensions supported by
this space.
- terms
- list all the indexed terms.
- docs
- list all the indexed documents.
- tdocs
- list all the indexed training documents.
compact filein fileout [flags]
compress a space by removing extraneous spaces or dimensions.
Reads space from filein and writes the compressed
space to fileout. flags controls what gets written to the
file. Possible flags include:
- -k=n
- set the number of dimensions of the space to n.
- --no-train
-
don't include the training document space.
- --no-doc
-
don't include the document space.
- --no-train-obs
-
don't include the training observation matrix.
- --no-obs
-
don't include the observation matrix.
search file k term
search for documents in file similar to a given term using
k dimensions. For more control over searching, use ssfind.
list file doc
display original terms used to index a document doc in the
semantic space file.
coords file [flags] [term | doc]
display semantic space coordinates of items. Available
flags are
- -k=n
- set the number of dimensions of the space to n.
- -docs
- include the document space.
- -tdocs
- include the training document space.
- -terms
- include the terms.
- -sf=n
- multiply the coordinates by a scale factor n.
If a term or document is specified after the flags then
only the coordinates for that term or document will be displayed,
otherwise the coordinates for all the items in the
selected spaces will be displayed.
odist file [flags] [term | doc]
display (Euclidean) distance of item from origin in the
semantic space. See coords to see how the flags work.
ssfind(1)
ssmake(1)
libSemanticSpace(3)
School of Electronics and Computer Science, University of Southampton
Jonathon Hare <jsh2@ecs.soton.ac.uk>
Table of Contents