The idea is use Word2Vec for find relations between words, map words into 2D dimensionality, analyze them for the collection of books about Harry Potter.
For installation please use conda. Just run:
make initAfter the installation was done successfully, activate your enviroment:
source activate hp2vecFor using conda and environments, please read full documentation of conda.
After activate source, run jupyter notebook hp2vec.ipynb. I saved pre-train model, so load it w2v.Word2Vec.load('models/500features_35minwords_10context') and have fun.
The hp2vec.ipynb has the precompiled code, so you can see some interesting relations and how the plots look like.