Transformer based Sentiment Analysis in Dravidian Languages

Arunaggiri Pandian Karunanidhi

Abstract:

The development of social media platforms have enabled users to express their thoughts and opinions
about entities freely, without any inadvertent implications it may have on a person/group. Due to the
volume of active social media users, it is becoming increasingly apparent for the need of automated
sentiment analysis systems for social media. This paper describes our work on the task of Sentiment
Analysis in Dravidian language-DravidianCodeMix 2021. We propose a soft voting classifier with the
help of other fine-tuned multilingual language models, achieving the best weighted F1-Score of 0.752,
0.619, and 0.648 in Malayalam, Tamil, and Kannada respectively. Our approach achieved the best results
in Tamil, securing 3
𝑟𝑑 rank in the language. The source codes of our systems are published