Researchers have developed STEPS, a novel Self-Triggered End-to-end Agentic Push Recommendation System, which has been fully deployed on Douyin, a platform with over 1 billion users. This system addresses limitations in traditional push notification strategies by reformulating the problem as a self-triggered agentic process. STEPS utilizes two decision transformer-based agents to dynamically decide when to send notifications and when to schedule its next invocation, aiming to balance real-time effectiveness with efficiency. Online A/B testing showed that STEPS increased user active days by 0.2843% and reduced the push permission disablement rate by 1.9089%, while a filtering agent cut computational overhead by 79.42%. AI
IMPACT This agentic system could influence how large platforms optimize user re-engagement and manage notification systems.
RANK_REASON Publication of a research paper detailing a new system with deployed results. [lever_c_demoted from research: ic=1 ai=1.0]
Read on arXiv cs.IR (Information Retrieval) →
- alphaXiv
- arXiv
- CatalyzeX Code Finder for Papers
- Connected Papers
- DagsHub
- Douyin
- Gotit.pub
- Hugging Face
- Litmaps
- ScienceCast
- scite Smart Citations
- STEPS
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
