Russian Propaganda
Analyzing the Russian Government's Use of Twitter
Russia's war in Ukraine has marked an inflection point for the future of the global order and democracy itself. Widely condemned for waging a war of aggression, the Russian government has sought to spread disinformation as justification for the war and to cause political division in the West. Seeking to garner insight into the efforts of the Kremlin in this regard, and Ben has studied the narratives, linguistic patterns, and social media strategies used by Russia to shift conversations about its invasion of Ukraine.
In a peer-reviewed conference paper published and presented at the 2nd ACM Workshop on Multimedia AI Against Disinformation, Ben showed how the Russian government has used its official Twitter accounts to shape English-language conversations about the war in Ukraine. 2,685 English-language tweets posted by 70 Russian government accounts between 1 September 2022 and 31 January 2023 were analyzed using a transformer-based topic model, with initial results showing the Russian government to portray itself as a noble world leader, while deflecting blame onto the “Kiev Regime” for starting the war. A semantic similarity analysis was then conducted to compare the narratives originating from Russian government Twitter accounts to 149,732 English-language tweets about the war in Ukraine to estimate these narratives’ spread. Results showed a segment of general discussion tweets to exhibit strongly similar language to Russian government tweets, but also highlight differences between the frequency and saliency of Russian government narratives.
In a follow-up "research in progress" paper published and presented at the 8th International Workshop on Social Media World Sensors, Ben examined the Kremlin's linguistic choices across social media, specifically analyzing the use of logical fallacies, among other malign linguistic methods. Logical fallacy detection has emerged as a novel and challenging task for language models, more complex than traditional fake news or hate speech detection. As part of this study, Ben introduced a curated dataset of tweets containing logical fallacies, RuFal, published by Russian government accounts about the war in Ukraine. Then, he tested a novel Entity-Aware Approach to logical fallacy detection—involving masking all named entities in the corpus. Initial results showed the Entity-Aware Approach to outperform baseline pre-trained language models by up to 3% when fine-tuned and tested on both a general dataset of logical fallacies and RuFal.
Ben's work contributes some of the first analysis of disinformation originating directly from official Russian government social media channels, providing insights that can help the foreign malign influence community to more effectively analyze the trove of Kremlin-authored social media content circulating online today.
In a peer-reviewed conference paper published and presented at the 2nd ACM Workshop on Multimedia AI Against Disinformation, Ben showed how the Russian government has used its official Twitter accounts to shape English-language conversations about the war in Ukraine. 2,685 English-language tweets posted by 70 Russian government accounts between 1 September 2022 and 31 January 2023 were analyzed using a transformer-based topic model, with initial results showing the Russian government to portray itself as a noble world leader, while deflecting blame onto the “Kiev Regime” for starting the war. A semantic similarity analysis was then conducted to compare the narratives originating from Russian government Twitter accounts to 149,732 English-language tweets about the war in Ukraine to estimate these narratives’ spread. Results showed a segment of general discussion tweets to exhibit strongly similar language to Russian government tweets, but also highlight differences between the frequency and saliency of Russian government narratives.
In a follow-up "research in progress" paper published and presented at the 8th International Workshop on Social Media World Sensors, Ben examined the Kremlin's linguistic choices across social media, specifically analyzing the use of logical fallacies, among other malign linguistic methods. Logical fallacy detection has emerged as a novel and challenging task for language models, more complex than traditional fake news or hate speech detection. As part of this study, Ben introduced a curated dataset of tweets containing logical fallacies, RuFal, published by Russian government accounts about the war in Ukraine. Then, he tested a novel Entity-Aware Approach to logical fallacy detection—involving masking all named entities in the corpus. Initial results showed the Entity-Aware Approach to outperform baseline pre-trained language models by up to 3% when fine-tuned and tested on both a general dataset of logical fallacies and RuFal.
Ben's work contributes some of the first analysis of disinformation originating directly from official Russian government social media channels, providing insights that can help the foreign malign influence community to more effectively analyze the trove of Kremlin-authored social media content circulating online today.
