Back

Analyzing Conspiratorial Content Across Singapore-Based Telegram Groups

Goyal, A.; Ligo, V. A. C.; Cheung, L. Y.; Lee, R. K.-W.; Saha, K.; Tandoc, E. C.; Kumar, N.

2025-07-17 public and global health
10.1101/2025.07.15.25331450 medRxiv
Show abstract

Telegram has emerged as a key platform for the circulation of conspiratorial narratives. We examine conspiratorial discourse within Singapore-based Telegram groups from 2021-2025. We analyze over 10 million words from three Telegram groups. We developed a logistic regression classifier to detect conspiratorial content, achieving an F1 score of 0.74 and expert-validated labeling accuracy of 72%. Topic models indicated dominant themes centered around elite control, vaccine risks, and globalist agendas. While most users rarely posted conspiratorial content, a small, highly active minority accounted for most of such messages. These users frequently forwarded messages across multiple groups, amplifying the spread of content with short but intense lifecycles (mean lifespan=6.8 days). Network analysis showed that users typically joined multiple groups in rapid succession and that conspiratorial messages traveled across groups within weeks. We underscore the importance of user-centric monitoring, time-sensitive interventions, and platform-specific models for content detection.

Matching journals

The top 2 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.