Banerjee, Somnath, juhendajaSharma, Rajesh, juhendajaAltmets, Anna-LiisaTartu Ülikool. Loodus- ja täppisteaduste valdkondTartu Ülikool. Arvutiteaduse instituut2023-11-022023-11-022023https://hdl.handle.net/10062/93926The COVID-19 pandemic not only cost human lives but also harmed industries like tourism which adds valuable contributions to the GDP of many countries. The pandemic affected global tourism in several ways, such as fewer flights, cancellations, lockdowns and restrictions, etc. This thesis studies COVID-19's impact on people's perception of travel safety leveraging sentiment analysis. Travel-related social media data was collected from Twitter and divided by the severity of the pandemic and the tweets volume of the regions to study the impact and patterns. For analysing data, a RoBERTa-base pretrained sentiment analysis model for tweets was employed. Sentiment scores over time were compared to understand the general trends. Although most of the tweets were neutral, there was an evident change in the proportion of negative tweets to positive. A word frequency was also verified during different periods in this work. Virus-related words were frequently used in positive and negative tweets. The study reveals that people cancelled or postponed their trips due to risks caused by the pandemic.engopenAccessAttribution-NonCommercial-NoDerivatives 4.0 Internationalmachine learningsentiment analysisnatural language processingCOVID-19magistritöödinformaatikainfotehnoloogiainformaticsinfotechnologyThesis