From Disinformation Dissection to Ideological Conflation
Abstract
”Magic-middle” social media influencers serve as critical intermediaries in contemporary information warfare, operating below systematic monitoring thresholds while maintaining trusted relationships within specialized niches. We present a mathematical framework for detecting coordinated influence operations through behavioural pattern analysis rather than content classification. Our approach mod-els influencer digital footprints using semantic embeddings processed through hierarchical stochastic processes: Hidden Markov Models for inter-topic transitions and Ornstein-Uhlenbeck processes for intra-topic semantic drift. This content-agnostic methodology enables politically neutral detection. The framework provides anomaly detection through ”cost of postage” metrics and structural comparison via Wasserstein distances between Gaussian processes.
