Towards Enhancement of the Lexicon Approach for Hindi Sentiment Analysis

Abstract

The amount of data from users in the form of reviews, comments and opinions in Hindi language is tremendously increasing on social media, blogs and online forums due to which sentiment analysis of Indian languages has turned out to be a predominant research area. Lexicon-based approaches are being used for analysing sentiments but they may not be as accurate as machine learning approaches. To enhance and improve the lexicon approach, a frequency score generation weighting scheme algorithm is proposed. Novel rules for handling some linguistic features such as negation, intensifiers are also implemented. Experiments performed to evaluate the performance of the proposed algorithm shows that the performance of lexicon-based sentiment analysis is increased significantly.