This study investigates the general public’s concerns about COVID-19 vaccination by their comments in social media (YouTube) with NLP techniques and time series analysis. A set of keywords are traced in order to better understand the changes in public opinion and responses at different stages of the pandemic, as well as the influences of fake news. These keywords were extracted from Macau netizens’ online comments based on word frequency, TF-IDF, and TextRank. It is observed that the misinformation dissipated abruptly after initiation of mass vaccination in Macau. We account for this change by the Prospect Theory. This study has shown that NLP techniques can assist in discourse analysis of people’s perceptions of COVID-19 vaccination, and people’s linguistic behaviours have been captured by the extracted keywords through text mining and time series analysis.