Bengali VADER: A Sentiment Analysis Approach Using Modified VADER

摘要

Sentiment analysis is an essential field of natural language processing (NLP) that classifies the opinion expressed in a text according to its polarity (e.g., positive, negative or neutral). Bengali NLP research is lagging behind English NLP, where there are very few works on Bengali sentiment analysis. In this paper, we approach this issue by modifying a popular English tool VADER to support Bengali sentiment polarity identification. We have compiled a Bengali polarity lexicon from the English polarity lexicon of VADER. Furthermore, we have modified the functionalities of English VADER, so that it can directly classify Bengali text sentiments without the requirement of Bengali to English translation using tools such as Google Translator, MyMemory Translator, etc. Our experiments demonstrate that the modified Bengali VADER significantly improves the sentiment analysis result of Bengali text over the current model.