Publication

Abstract : Sentiment analysis is a valuable method for analyzing texts and understanding the opinions, attitudes, and emotions expressed toward different subjects. Its application to large-scale data, particularly on social media platforms like Twitter or Facebook, offers valuable insights. However, Twitter data presents unique challenges due to its complex and noisy nature, as well as its intricate syntactic and semantic structures. Moreover, analyzing sentiment becomes even more difficult when dealing with multimodal Twitter data in Indian languages. In order to address these obstacles, it is essential to create a robust framework capable of effectively processing Twitter content across diverse languages and encodings. To ascertain the most optimal methodology, we carry out extensive assessments employing multiple variations of multilingual and single-language models, ultimately relying on the evaluation outcomes to guide our selection process. Our aim is to create a comprehensive model capable of effectively characterizing all aspects of Twitter data. Through a comprehensive examination of the data, we gain a deeper understanding of the sentiments expressed across various languages used on Twitter in India. These results offer valuable insights into the diverse range of emotions, opinions, and attitudes expressed by Indian Twitter users, enabling a more nuanced understanding of public sentiment in this multilingual context.