Contextual Lexicon Based Sentiment Analysis in Myanmar Text Reviews

A lot of information related to several commercial application available online which can be used to provide the guidance and suggestions to possible new customers. People desire to distribute the opinions and state the sentiments in their own language. Sentiment analyzers developed for English language, are not workable for Myanmar language. Mining sentiments in Myanmar text come with a lot of issues and challenges. The direction of the sentiment is highly depend on the context of sentiment text. Thus, it is significant challenge to consider contextual lexical information in order to correctly classify the polarity. This paper aims to improve the existing challenges problem of language and analyze the sentiment classification of food and restaurants domain by using contextual analysis with lexicon based approach in Myanmar text reviews. The effect of intensifier, negations and objective words are important role in the context of sentiment orientation. This paper addresses sentiment classification for Myanmar Language and overcome one of the problems of language specific challenges. The accuracy of the proposed system is higher than the classification without using context information (negation, intensifier and objective words). Overall accuracy of the proposed system is 92% and weighted average F-measure for imbalance class of 1200 reviews is 0.93.