In this repo, I am analyzing rule based sentiment analysis systems and ML based sentiment analysis systems. The 10000 Movies review from the Cornell Dataset will be used in the sentiment analysis. Rule Based Sentiment Analysis systems will include VADER and Sentiwordnet. The ML Based system will include Naive Bayes Classifier algorithm.
Rule based sentiment analysis with VADER achieved :
-
True Positives Accuracy : 69.5%
-
True Negatives Accuracy : 42.2%
ML based sentiment analysis with Naive Bayes achieved :
-
True Positives Accuracy : 73.4%
-
True Negative Accuracy : 23%
Pull requests are welcome