#####################################################
#Computational Linguistics 18-19
#DISSECT tutorial
#How to: Get ready to play!
#sandro.pezzelle@unitn.it
#####################################################

#requirements:
# 1. terminal
# 2. python
# 3. 10-15 minutes

#In what follows we will create a virtual environment with
#everything set up for our hands-on session :)
#the nicest thing of virtual environments is that
#they can be removed with sudo rm -r <virtualenv-folder> in case you
#do not need to work with semantic spaces anymore!
#OPEN THE TERMINAL and copy-paste the content of this txt file,
#press Enter and wait for the magic happening!

#install virtual env
cd ~
sudo pip install virtualenv

#if that does not work, check if you have pip installed...
#sudo easy_install pip
#cd ~
#sudo pip install virtualenv

#make a directory called dissect-env
cd ~
virtualenv --python=/usr/bin/python2.7 dissect-env

#activate virtual environment
source dissect-env/bin/activate

#open folder
cd dissect-env/

#be sure you have NumPy, SciPy, etc. installed!
pip install numpy
pip install scipy
pip install cython
pip install sparsesvd

#download dissect and install it
git clone https://github.com/composes-toolkit/dissect
cd dissect
python2.7 -m setup.py install
#if this does not work
#try instead:
#sudo python2.7 -m setup.py install
#and type your SUDO password
#you should have DISSECT installed at this point!

#now let's install gensim, a useful library for playing with
#and building distributional semantics models
cd ~
cd dissect-env/
easy_install -U gensim
#now you have gensim installed!

#Let's download some useful data...
mkdir class
cd class/
curl -L -o 'class-stuff' 'https://sandropezzelle.github.io/Other/class-stuff.zip'
unzip 'class-stuff.zip'
cd class-stuff

#get a .pkl version of the "predict" DSM
python2.7 bin-to-pkl.py


#OPTIONAL begins
#get a small but "real" corpus
curl -L -o 'text8.zip' 'http://mattmahoney.net/dc/text8.zip'
unzip 'text8.zip'
#count cooccurrences, occurrences of corpus (NOTE: it takes time and memory!) + extract contexts
python co-occ.py
sh extract-cols.sh
#build a "count" DSM based on cooccurrences, occurrences counts
python build-count-model.py
#OPTIONAL ends


#if you get ImportError: No module named composes.semantic_space.space
#modify bin-to-pkl.py changing the path of vectors.bin
#and try instead:
#cd ~
#cd dissect-env/dissect/src
#python2.7 ../../class/class-stuff/bin-to-pkl.py


#When everything is properly set up:
deactivate
cd ~

#you're done! :)