COIMBRA, M. V. S.; http://lattes.cnpq.br/6379435876952280; COIMBRA, Mariana Victória Souza.
Résumé:
The present article involves an analysis of sentiments and emotions expressed in Tweets related to the Ukraine War, through an examination of topics discussed by users on the Twitter platform. This study aims to comprehend how users react to the ongoing event, which aspects of the war people are discussing on the platform, and how they feel about this occurrence. Additionally, it seeks to identify correlations among variables present in the Tweets, such as location, user profile information of the post author, and the nature of their opinions. These analyses were conducted through natural language processing tasks, including exploratory data analysis and the application of sentiment classifiers for Tweets using pre-trained data models. The analyzed data includes Tweets collected from the beginning of the conflict, which started in February 2022, until October 2023, gathered through hashtags related to the War. For sentiment and emotion analyses, the RoBERTa variant was employed. Tweets were categorized into sentiments such as positive, negative, or neutral, and emotions including joy, anger, fear, disgust, optimism, pessimism, surprise, and love. The results indicated that the majority of English tweets express anger and anticipation as predominant emotions, with negative and neutral sentiments having greater prevalence, exceeding 50% of the analyzed sample. Some of the most recurrent phrases in the analysis reference support for Ukraine and call for an end to the war. Similarly, phrases expressing concern about the crisis, weapons, and fatalities are common. In most posts, individuals demonstrate concern about the armed conflict and express support for Ukraine. Future work could incorporate more tweets to encompass the analysis and visualize the correlation of additional attributes related to posts, such as engagements and likes.