GRAMMAR RULE BASED SENTIMENT ANALYSIS TECHNIQUES FOR TAMIL TWEETS CLASSIFICATION

Abstract

The advent of social media has allowed people to express their sentiment about a product, service or movies etc., Sentiment is a view or attitude of a user towards any topic, object, event or service. In general, sentiment has always been influencing people s decision making. Sentiment analysis in natural language has been intensively researched over the recent years, but there are still many issues to be addressed. One of the main problems is the lack of resources to carry out precise sentiment classification. Most of the research on sentiment classification deals with the polarity classification problem, although for many practical applications this rather coarse sentiment measure is not sufficient. Therefore, it is important to design methods which are able to perform precise sentiment classification in natural language. newlineThe major goal of this research is to develop a sentiment framework for Tamil tweets genre classification based on grammar rules. In this thesis, three key challenges affecting the sentiment classification in Tamil tweets are delineated and the possible methods to address these problems are proposed. First, informal nature of tweets has been found critical to the sentiment classification performance. Therefore, a combination of grammar rules based on adjectives and negations is proposed to classify Tamil tweets precisely. Second, the people often use slang words, abbreviations and mixed words to express their sentiments. The proposed system uses a method called domain-specific tags to incorporate non-grammatical words such as slang words and mixed words etc. Third, the morphological richness of the Tamil language leads to performance degradation when tweet length is more complex. To overcome this, the proposed grammar rules are incorporated into N-gram approach and machine learning methods for better performance. newlineIn this thesis, the proposed methods are grouped into three different approaches based on their functionality; one that uses syntactic functionality of words to predict Tamil tweets