LLM-Enhanced Topical Trend Detection at Snapchat
Hangqi Zhao, Jay Li, Abhiruchi Bhattacharya, Cong Ni, Jason Yeung + 4 more
TLDR
This paper introduces Snapchat's new LLM-enhanced system for detecting emerging topical trends on its short-video platform at production scale.
Key contributions
- Integrates multimodal topic extraction, time-series burst detection, and LLM-based consolidation.
- First published end-to-end system for topical trend detection on short-video platforms at production scale.
- Achieved high precision in identifying meaningful trends through six months of continuous human evaluation.
- Deployed globally, improving content freshness and user experience in content ranking and search.
Why it matters
This paper presents a novel, production-scale system for detecting topical trends on short-video platforms using LLMs. Its successful deployment at Snapchat demonstrates significant improvements in content freshness and user experience, setting a new standard for trend detection in dynamic social media environments.
Original Abstract
Automatic detection of topical trends at scale is both challenging and essential for maintaining a dynamic content ecosystem on social media platforms. In this work, we present a large-scale system for identifying emerging topical trends on Snapchat, one of the world's largest short-video social platforms. Our system integrates multimodal topic extraction, time-series burst detection, and LLM-based consolidation and enrichment to enable accurate and timely trend discovery. To the best of our knowledge, this is the first published end-to-end system for topical trend detection on short-video platforms at production scale. Continuous offline human evaluation over six months demonstrates high precision in identifying meaningful trends. The system has been deployed in production at global scale and applied to downstream surfaces including content ranking and search, driving measurable improvements in content freshness and user experience.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.