Prepare your corpus file and put it in the LSA/bin/debug directory
Run the LSA executable and follow the onscreen instructions
The octave part requires that you have octave installed. You need octave >= 3.0. Make sure you have both octave3.0 and octave3.0-headers,e.g. for ubuntu:
sudo apt-get octave3.0 octave3.0-headers
You also need to have Arpack installed
sudo apt-get install libsuitesparse-dev libarpack2-dev
You also need to install the arpack from octave forge.
This is as easy as downloading the arpack package, opening a octave shell in that directory, and typing 'pkg install arpack-1.0.8.tar.gz'
When the LSA executable prompts you to run the octave script, you can type 'octave svds_oct $corpus_name.SPARSE $300' where $corpus_name should be replaced by the name of your corpus and $300 by the number of dimensions you want.
The process will create an LSA space object, serialized to file, that you can load and use. GnuTutor expects this file to be in the Databases directory. You may also need to change the config file if you decide you want a special name for your new space.