Assamese VADER: A Sentiment Analysis Approach Using Modified VADER

Sentiment Analysis is a Natural Language Processing (NLP) technique that determines the opinion towards an entity, identifying the opinion as positive, negative, or neutral. Extensive research has taken place for high-resourced Languages like English, whereas for Low resourced Indo-Aryan languages, it is still an area in progress. This paper attempts to perform Sentiment Analysis on Assamese Texts, which is a morphologically rich yet Low-resource Indo-Aryan Language, using the concepts of popular sentiment analyzer named “Vader”, while taking “Bengali-Vader” as its backbone. The process follows that of a traditional Vader tool, with consideration towards creating a dictionary of negative booster words and creating an Assamese Lexicon, pre-processing of data, boosting the valence of each word, valence calculation and sentiment categorization of text based on the valence. The necessity of proper dataset in this method is immense, and even though the lack of good translation tools and resources to feed the model, this model was experimented on a set of Assamese texts from a renowned Assamese novel. The comparison was done by manually translating the Assamese sentences to its Bengali and English forms and it shows significant results when compared with the Bengali and English counterparts.