When War is Not a “War”: Analyzing Russian State-Sponsored Online Propaganda

• In Event: Public Opinion and Political Trust in Post-Communist Countries

Abstract

While the rise of social media has positively contributed to democratic discourse worldwide, over the years, it has also become a platform for state-sponsored propaganda aimed at shaping public opinion in support of incumbent authoritarian (mostly dictatorial) regimes. Distinctively, Russian and Chinese governments have utilized various online platforms to (a) disseminate disinformation and propaganda to create “manufactured consent” among the public for its domestic and foreign policy objectives; (b) suppress pro-democratic voices through systemic restrictions; and (c) conduct mass surveillance of citizens. For instance, at the onset of Russian aggression into Ukraine in February 2022, the Russian government introduced a law on “fake news” that effectively prohibited the use of the word “war”, the usage of which is now punishable for up to 10 years in prison. Due to this legislature, dozens of news platforms were shut down and thousands of Russians were criminally charged. The scale and scope of online state-sponsored propaganda have reached unprecedented volumes just in the last 12 months.

Russian propaganda dissemination is not limited to a domestic audience. Instead, the official Kremlin’s narratives about justifying “Russian special military operation” (in lieu of using the word “war”) in Ukraine have been deliberately targeted at all Russian-speaking former Soviet republics, particularly Central Asia. What is particularly striking is that these narratives often incite extremism and violence. Human behavior is surely the result of exposure and feedback from the cultural, social, and physical environment. Such commonplace learning is sensitive to the accrual of experience over time, and so it would seem for exposure to mis/disinformation online. Preventing our understanding of the relationship between behavior and mis/disinformation/extremist content is the definition of such content, typically ambiguous and subjectively established.

In this project, we ask: To what extent does foreign state-sponsored online propaganda that employs radicalization and extremism narratives as a political weapon succeed in persuasion and recruitment? By engaging this central question, we attempt to investigate the online persuasion dynamics used by the Russian government in the transnational context.
We draw from the system-supportive manufactured consent theoretical framework developed by Herman and Chomsky (1988) who frame mass media platforms as “effective and powerful ideological institutions that carry out a system-supportive propaganda function… without overt coercion”.

We collect and analyze data from Russian-speaking Telegram and WhatsApp channels and related web blogs. These channels have been identified with the assistance of local scholars/researchers in Central Asia. Moreover, offline events documented by local observers are also analyzed. We utilize a mixed method approach, including qualitative and quantitative analysis of online and offline content to extract the relationships between online and offline settings.
We begin with a qualitative analysis of this online multimodal content for identifying themes, and possible reasons for toxic/hateful behavior while building a codebook for these respective components. Further, we utilize public multimodal toxicity hate speech datasets (Gomez et al. 2020; Kiela et al. 2021) for training a machine learning model that measures the level of toxicity & hateful language through regression analysis, followed by human validation. For multimodal learning, we will utilize transfer learning approaches to fine-tune existing language (for text) and image models to generate more contextually cohesive representations. We use bi-directional transformers (Kiela et al. 2020), and transformer-based language and image models along with knowledge representations utilizing DGL-KE for KG embedding models (e.g., node2vec, TransE, etc.), allowing us to disambiguate the subjective/ambiguous concepts.

By exploring the scope and themes of influential propaganda campaigns by Russia, we aim to (a) identify and classify highly influential actors who are actively producing and spreading Russian propaganda; (b) identify and classify groups and entities who are deliberately targeted; (c) analyze network linkages of content creators (i.e., propagandists); (d) explore major narrative themes within disinformation content; (e) the quantitative analysis of the online content for measuring toxicity and hate produced by these state-sponsored propaganda materials. We will present our preliminary findings from this ongoing research.

Authors

• Dilshod Achilov, University of Massachusetts - Dartmouth

• Ugur Kursuncu, University of South Carolina