Sentiment Analysis in Twitter for NRC Hashtag Emotion

ABSTRACT Sentiment Analysis (SA) is the most studied field now a days, is also known as Opinion Mining (OP). SA is a task of analyzing people's thoughts, point of view, attitude, etc. towards particular product, topic, organization, etc. As electronic commerce (ecommerce) is increasing rapidly, people's review on the e-commerce website is also increasing rapidly. Twitter is one of the most used microblogging websites these days. Detecting emotions from twitter posts automatically is a challenging task because informal nature of language is used. As great attention is given to polarity of words i.e. positive or negative, here I am using NRC word emotion association lexicon which have ten emotional categories. As NRC-10 didn't include expressions used every day in social media, here expansion of NRC word - emotion association lexicon is done for language used in social media. Expansion is done using multi-label classification and word level features are extracted from unlabeled tweets. Results show great improvement after expansion of lexicons. Keywords: Sentiment Analysis, NRC-10, WEKA, MEKA, Tweets, anger, anticipation, disgust, fear, joy, sadness, surprise, trust