Exploring Word Embedding for Bangla Sentiment Analysis
Sakhawat Hosain Sumit, Md. Zakir Hossan, Tareq Al Muntasir, Tanvir Sourov · 2018
Sentiment Analysis (SA), sometimes known as opinion mining, polarity analysis or emotional AI, is a study of analyzing user's reviews, ratings, recommendations and other forms of online expressions. Most of the research work on SA in Natural Language Processing (NLP) are focused on the English language. However, Bengali is spoken as the first language by almost 230 million people worldwide, 163.9 million of whom are Bangladeshi. These people are found to get increasingly involved in online activities on popular microblogging and social networking sites, sharing opinions and thoughts and most of them are in Bengali and Romanized Bengali (English character to write Bengali) language. These online opinions are changing the way of doing business. And lots of data are being generated each year which are being underutilized. In this paper, we have experimented current state of the art word embedding methods Word2vec Skip-Gram and Continuous Bag of Words with an addition Word to Index model for SA in Bangla language. Word2vec Skip-Gram model outperformed other models and achieved 83.79% accuracy.