ChestyBot: Detecting and Disrupting Chinese Communist Party Influence Stratagems

TL;DR

The study addresses real-time detection and disruption of foreign malign influence campaigns on social media. It introduces ChestyBot, a pragmatics-focused CNN-based NLP model trained on four influence stratagems (inform, invoke, deflect, recast) and built atop a snowball-sampled network of CCP/PLA-linked accounts, with echo chambers identified via the Louvain algorithm. The approach yields a high validation accuracy (98–99%) and demonstrates practical early-detection potential, identifying hundreds of propaganda tweets and mapping liminal nodes for targeted disruption. By integrating automated detection with network-topology insights, the paper argues for a proactive defense in the information domain that can guide counter-messaging and moderation to curb emerging influence campaigns.

Abstract

Foreign information operations conducted by Russian and Chinese actors exploit the United States' permissive information environment. These campaigns threaten democratic institutions and the broader Westphalian model. Yet, existing detection and mitigation strategies often fail to identify active information campaigns in real time. This paper introduces ChestyBot, a pragmatics-based language model that detects unlabeled foreign malign influence tweets with up to 98.34% accuracy. The model supports a novel framework to disrupt foreign influence operations in their formative stages.

Figures (6)

• Figure 1: Each colored box represents a decision that can be influenced by the application of Stratagems. The introduction of stimuli prompts the actor to make a decision beneficial to an influencer’s selected course of action [Starting with decision I2]. The deliberate application of Stratagems directs decisions toward a specific outcome [Green Arrow to A5] and, failing that, accounts for the range of possible decisions that would lead to acceptable managed outcomes [Green Boxes]. Adapted from communications with Colonel Drew Cukor, USMC in 2019.
• Figure 2: Our main effort is to make the external measurability of informational artifacts easier, which is a problem well described in a 2015 U.S. Special Operations Command publication command2015operating.
• Figure 3: We used the Snowball Sampling Method to build Set1.
• Figure 4: The NATO-funded taxonomy kelly_paul_2020 depicted here provides a digest of Russian influence stratagems used during the annexation of Crimea.
• Figure 5: Training for ChestyBot