Exploring Crisis-Driven Social Media Patterns: A Twitter Dataset of Usage During the Russo-Ukrainian War

On 24 February 2022, Russia’s invasion of Ukraine, now known as the Russo-Ukrainian War, sparked extensive discussions on Online Social Networks (OSN). We initiate a data collection using the Twitter API to capture this dynamic environment. Next, we perform an analysis of the topics discussed and a detection of potential malicious activities. Our dataset consists of 127.2 million tweets originating from 10.9 million users. Given the dataset’s diverse linguistic composition and the absence of labeled data, we approach it as a zero-shot learning problem, employing various techniques that require no prior supervised training on the dataset.

Our research covers several areas, including sentiment analysis capturing the public’s response to the distressing events of the war, topic analysis comparing narratives between social networks and traditional media, and examination of the correlation between message toxicity levels and Twitter suspensions. Furthermore, we explore the potential exploitation of social networks to acquire military-related information by belligerents, presenting a pipeline to classify such communications.

The findings of this study provide fresh insights into the role of social media during conflicts, with broad implications for policy, security, and information dissemination. Finally, due to the recent Twitter API changes, we share anonymized data for any further research purposes.