Analyzing Emotions and Modeling Discussed Topics Among Platform X Users Regarding the Russia-Ukraine War by ChatGPT

With the onset of Russia's attack on Ukraine on February 24, 2022, people worldwide shared their opinions and emotions about this event on social networks. This study, utilizing Tweet analysis, aims to extract highly debated topics and user sentiments about the Ukraine war among users of platform X. Extracting these opinions can be highly beneficial for decision-makers in various countries. This research faces three primary challenges: the first challenge being that sentiment analysis methods are more effective on a broad scale but lose accuracy when applied within a specific domain. Additionally, determining the number and label of clusters in topic modeling is also a fundamental challenge. To address these challenges, we propose an improved algorithm based on VADER for sentiment analysis. Furthermore, a two-stage algorithm using coherence and overlap criteria is proposed to determine the number of discussed topics' clusters, followed by employing a chat GPT-based approach for labeling the clusters. The results indicate that negative sentiments are more than twice as prevalent as positive sentiments, and the significant debated topics among users are divided into three clusters named “Transformations,” “Stability,” and “Unity.”