An unsupervised Bayesian inference method with amortized variational inference detects coordinated inauthentic accounts on Twitter by clustering on account-level traits and shared narratives, substantially outperforming naive baselines and approaching supervised performance.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SI 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Unsupervised detection of coordinated information operations in the wild
An unsupervised Bayesian inference method with amortized variational inference detects coordinated inauthentic accounts on Twitter by clustering on account-level traits and shared narratives, substantially outperforming naive baselines and approaching supervised performance.