TwitterHybridClassifier

• Authors: L V Raju Nadimpalli, Akhil Sharma
• Date: 16/05/17

These python scripts provide the TwitterClassifier I used for my B.tech final year project

Dependencies :

• python 2.7 or a python 3.4 versions
• Natural Language Toolkit ( NLTK - http://www.nltk.org/ )
• scikit-learn ( Machine Learning in Python - http://scikit-learn.org/stable/ )
• ark_twitter_nlp Pos tagger ( http://www.cs.cmu.edu/~ark/TweetNLP/ )

I also used ark_twitter_nlp Pos tagger as well as the NLTK pos tagger.

The lexicon this library uses are:

• NRC Hashtag Sentiment Lexicon http://www.umiacs.umd.edu/~saif/WebPages/Abstracts/NRC-SentimentAnalysis.htm

• Liu's Opinion Lexicon http://www.cs.uic.edu/~liub/FBS/sentiment-analysis.html

Running the code :

• To obtain the end output make an instance of the TwitterHybridClassifier() class providing your pos tagged and tokenzied dataset as arguments

Output :

The final output of would be the sentiment(i.e. 'positive', 'negative' or 'neutral') predicted by the classifer along with the accuracy and confusion matrix.

Instead of using the NLTK tokenizer I have used the tokenizer built by Brendan O'Connor(http://github.com/brendano/tweetmotif)

The datasets I have used in my codes are of SemEval 2014 which I cannot provide because of legal reasons. You can contact the organisers of the SemEval for the datasets.