Cross-Platform Framework Detects Synthetic Political Narratives
A new technique for identifying misleading political narratives on social media has been developed by researchers and published in arXiv:2605.21540. This innovative method employs four distinct metrics: lexical diversity, temporal burstiness, rhetorical repetition, and semantic homogenization, culminating in a score known as the Synthetic Narrative Coordination Score (SNC). The study analyzed over 353,000 posts from six channels on Telegram and nine communities on Reddit, focusing on six geopolitical events projected from 2023 to 2026. Notably, the channel IntelSlava exhibited the lowest lexical diversity and ranked highest in the SNC score for four out of the six analyzed events.
Key facts
- Framework uses four coordination signals: D(C), B(C), R(C), H(C)
- Combined into Synthetic Narrative Coordination Score SNC(C)
- Applied to 353,223 records from six Telegram channels and nine Reddit communities
- Covers six geopolitical event windows from 2023 to 2026
- IntelSlava shows lowest lexical diversity (MATTR 0.52–0.54)
- IntelSlava shows highest burstiness (B=+0.48 to +0.73)
- IntelSlava shows highest rhetorical overlap (Jaccard 0.12)
- IntelSlava ranks first in SNC(C) on four of six event windows
Entities
Institutions
- arXiv
- Telegram
- IntelSlava