Emotions in the time of COVID-19: A sentiment analysis of tweets during the nationwide lockdown in India

BackgroundCOVID-19 pandemic is unprecedented in terms of burden, nature and quantum of control measures and public reactions. We report trends in public emotions and sentiments before and during the nation-wide lockdown implemented since 25th March 2020 in India. MethodsWe collected a sample of tweets containing the keywords coronavirus or COVID-19 published between 12th March and 14th April in India. After pre-processing, the tweets were subjected to sentiment analysis using natural language processing algorithms. ResultsOur analysis of 226170 tweets revealed a positive public sentiment (mean sentiment score=0.25). Tweets expressing a given sentiment showed significant (p<0.001) waning of negativity; negative tweets decreased (39.3% to 35.9%) and positive tweets increased (49.8% to 51.8%). Trust (0.85 words/tweet/day) and fear (0.66 words/tweet/day) were the dominant positive and negative emotions, respectively. ConclusionsPositive sentiments dominated during the COVID-19 lockdown in India. A surveillance system monitoring public sentiments on public health interventions for COVID-19 should be established.