Springer’s Social Network Analysis and Mining (SNAM), a prestigious journal, publishes groundbreaking research at the intersection of computational and information sciences and the social sciences. Recently, SNAM published one of our studies titled “Modeling cross-platform narrative templates: a temporal knowledge graph approach.” The authors present a novel framework that uses temporal knowledge graphs to model relationships among narratives across social media platforms.

The study presents an innovative method to model the relationships between narratives on social media and detect narrative flow templates—temporal sequences of how content disseminates across platforms. It addresses two major challenges: representing narrative relationships for macro-level insight and identifying recurring narrative flows used by different ideological actors. 

The analysis reveals that Pro-Taiwan and Pro-China information actors leverage distinct narrative flow templates—temporal sequences of content spreading across platforms—to disseminate their messages. Rare narrative patterns with high confidence and frequent ones with high support and confidence (e.g., TikTok → YouTube → Instagram) were uncovered using sequential mining techniques.

The research shows that temporal knowledge graphs combined with AI-driven narrative extraction and sequential pattern mining offer an effective model to uncover how narratives evolve and spread across social platforms. This is particularly relevant in understanding coordinated information campaigns, such as those observed in Taiwan recently. 

Prof. Agarwal said, “By identifying how narratives move through platforms, we gain deeper insight into the cross-platform communication strategies of information actors and can better inform countermeasures for digital influence campaigns.”

Click here to read the full article.