Topical Sentiment Classification to Unmask the Concerns of General Public during COVID-19 Pandemic using Indian Tweets

Abstract

One of the vibrant social media platforms is which has more than half a million uses across the globe. It has become a popular means for dissemination of the news, to discuss on world events. It is also medium to converse about health centric information with updates given by the concerned officials and general public health-related information, during an abnormal situation like COVID - 19 pandemics. As the dimension of data and the linguistics of data been discussed is diverse in nature, it is a challenging task to identify only the content that is interesting and useful. Few studies have been done exploring the regional languages than other English. In this work, we explored huge number of tweets on post-lock down during Covid-19 pandemic by analyzing the sentiments expressed on the tweets and topic identification. To do the same we have employed English 2,126,421 and 76,265 Tamil tweets for analyzing and discussing the usefulness of sentiment analysis and topic modeling in both of the languages. Seven subjects were that are ranked on the analysis of content discussed from India, in Twitter during the four months from May 2020 and August 2020. Tamil tweets are investigated to understand the sentiments of people during anamount of time and the association to the media information published, and the assessment of the psychological behavior of human in India. It is always significant to understand the human opinions, information communication and building an agreement including social media in different regions of the nation.