brijml/mikolov_word2vec

Distributed Representations of Words and Phrases and their Compositionality

mikolov_word2vec

This project is besed on the paper "Distributed Representations of Words and Phrases and their Compositionality" by Tomas Mikolov et al.

1. Create a virtual environment using anaconda(install anaconda2 if you do not have it installed)

   $conda create -n <env-name> python=2
   

2. Activate the virtual environment

   $source activate <env-name>
   

3. Install the required package

   $conda install --file requirements.txt
   

4. Install the corpus using nltk download

   $ipython
   	>>>import nltk
   	>>>nltk.download()
   

5. Run the scipt word2vec.py to find the word representations

   $python word2vec.py
   

6. The word representations are stored as dictionary where each key-value pair is a word(string) and its vector representation (numpy arrray) which is stored as a pickle file.