The occurrence of the novel coronavirus has changed a series of aspects related to people's everyday life, the negative effects being felt all around the world. In this context, the production of a vaccine in a short period of time has been of great importance. On the other hand, obtaining a vaccine in such a short time has increased vaccine hesitancy and has activated anti-vaccination speeches. In this context, the aim of the paper is to analyze the dynamics of public opinion on Twitter in the first month after the start of the vaccination process in the UK, with a focus on COVID-19 vaccine hesitancy messages. For this purpose, a dataset containing 5,030,866 tweets in English was collected from Twitter between 8 December 2020-7 January 2021. A stance analysis was conducted after comparing several classical machine learning and deep learning algorithms. The tweets associated to COVID-19 vaccination hesitancy were examined in connection with the major events in the analyzed period, while the main discussion topics were determined using hashtags, n-grams and latent Dirichlet allocation. The results of the study can help the interested parties better address the COVID-19 vaccine hesitancy concerns.
Keywords: COVID-19 vaccination; natural language processing; opinion mining; stance analysis; vaccine; vaccine hesitancy.