Using Word2Vec for sentiment analysis

• Python 2.7
• Check requirements in requirements.txt

pip install -r requirements.txt

• If installation fails, or errors occur when running the application, following this to install virtualenv and install the requirements in the virtual environment. Recommended

• Directory structure and data preparation
  • create a directory called save in root
  • if you want to run similarity.py, download the google news corpus from this, and put it to directory dataset
• Run the following application scripts, based on your interest.

• wordvector.py: uses Word2Vec to train and compute word vectors for our corpus
  • The corpus file is in dataset/all
  • The model is saved in save/model.bin
• docvector.py: extracts feature vectors using four algorithms, and save them to directory save
  • there four algorithms are:
    • Word Averaging
    • Word Averaging + SentiWordNet
    • NLPF (Feature sets extracted in assignment 5.4)
    • NLPF + SentiWordNet
    • For each algorithms, six feature vectors are produced, as we are to perform 3-fold cross validation, and we have a feature set and a test set for each
• classification.py: apply three machine learning classifiers to the feature vectors produced by docvector.py and print performance metrics for each combination of classifier and feature vector
  • Random Forest
  • SVM
  • Naive Bayes
• similarity.py: uses the word vectors computed by wordvector.py, and predefined vectors trained from Google News Dataset to find the 10 most similar words for the following keywords:
  • apple
  • iphone
  • ipad
  • ipod
  • mac
  • macbook