Affect Analysis Using Social Media Multilingual Textual Data To Predict Voting Behaviour

Abstract

newline People use social media not only for personal communications with newlinefriends and family but also to voice their opinion on events or newlinehappenings which take place in society, country or the world. Social newlinemedia has made the entire world a global village. Social media provides newlinea platform to its users to mark their presence and value their opinion in newlinesociety. Not only individuals but business organizations, government newlineetc. also use social media to reach out to their targeted users. As newlinesupport of only global language from social media hampers the growth newlineof social media users, social media supports regional languages of newlinevarious countries. Thus a huge amount of content in multiple languages newlineis getting generated by social media users every day. newlineThere are a variety of use cases of social media from marketing to newlinecampaign. Election campaign is one such use case which has proved newlineits effectiveness since a decade. Social media users post their views newlineand comments on social media in reaction to the online election newlinecampaign. Thus it opens up the opportunity to mine this user-generated newlinecontent to predict votersand#8223; inclination towards political parties or newlinecandidates. Twitter is one of the social media which has been used for newlinethe election campaign for the Indian general election of the year 2014 newlineand the year 2019. newlineSentiment analysis is a computational technique which has been widely newlineused to extract and understand social media usersand#8223; opinion by newlinecategorizing text into positive or negative classes. Affect analysis newlineprovides more in-depth insight compared to sentiment analysis which newlinecategorizes text into multiple emotion categories like joy, sadness, fear newlinev newlineetc. This research study has performed affect analysis of multilingual newlinetweets to predict voting behaviour. Three languages, namely English, newlineHindi and Gujarati, were taken into consideration for multilingual tweets. newlineTwo major political parties of India, namely Bharatiya Janata Party (BJP) newlineand Indian National Congress were considered for this research study. newlineMultilingual tweets