Sentiment Analysis of Movie Reviews in Regional Language: Kannada

Sentiment analysis is the process of gathering and examining people's thoughts, feelings, attitudes, views, etc. about various subjects, goods, and services. The rapid development of applications such as YonTube, Instagram, Twitter and Facebook, and web pages encourages people to create a vast amount of reviews and comments regarding daily activities, goods, and services. Businesses, governments and researchers can use sentiment analysis as a potential tool to gather and examine public opinion and mood in order to obtain business intelligence and improve decision-making. Kannada language has been taken here as resources are very less on the internet. In this study, Movie reviews were collected in English language from IMDb website and translated into Kannada using deep translator for translation. First emojis and punctuation are removed then text is tokenized and sent to 2 BiLSTM (Bidirectional Long Short-Term Memory) layers. Then they are fed into many Fully Connected layers and a softmax activation function. The proposed model is compared with other deep learning models such as BERT (Bidirectional Encoder Representations from Trans formers) and DistilBERT. The performance of the proposed model is analyzed in terms of accuracy. Considering the accuracies of the models being BERT (0.616), DistilBERT (0.67) and Multilingual Representations for Indian Languages(MuRIL) (0.690), MuRIL was a better choice building the base model for the study than the other two counterparts.