Sentiment Analysis: An Approach for Analysing Tamil Movie Reviews Using Tamil Tweets

Abstract

Indian languages are often used in social media messages. Tamil is one of the oldest languages which has been used often in tweets. Sentiment Analysis (SA) is a multidisciplinary unit which is a part of text mining as well as natural language processing.  Sentiment analysis has gained incredible development in recent times mostly for English language. However very less work of sentiment analysis has done for Indian languages like Hindi, Tamil, Kannada etc., In this chapter, Tamil tweets are focused to find the sentiment of the Tamil movie reviews. It is essential to analyze the Tamil language content for tweets and get perception of opinion expressed by the tweets. The objective is to classify the sentiment of the Tamil movies based on Tamil tweets using Tamil SentiWordNet (TSWN). Term Frequency - Inverse Document Frequency (TF-IDF) method is proposed to find the sentiment polarity of the Tamil movie dataset. Domain specific ontology is applied to identify the primary sentiment categorization of the Tamil movies. In contextual semantic, the sentiment of a word may flip based on the neighbouring word. In this research, sentiment-bearing terms and its neighbouring terms in Tamil tweets are evaluated using contextual semantic sentiment analysis to get more accurate result for the movie sentimental classification.