Twitter Sentiment Analysis

There are three separate scripts:

1. word2vec_sentiment,py uses a pre-trained word2vec model

2. lsa_sentiment.py uses simple bag-of-words model with LSA

3. joint_sentiments.py combines both 1 and 2

Notes:

- Removing stop-words didn't improve results much

- Combining two models improved results slightly but
	nothing radical

- The results are little skewed since no. of samples 
	for all three classes are not equal.

Additional Requirements:

The pre-trained word2vec model is available to download at below link with relevant literature, and once it downloaded it must be placed in 'models/'' sub-directory.

http://www.fredericgodin.com/software/