Researchers have developed MORE (Multi-task cO-evolving Ranking modEl), a novel architecture that integrates multi-task learning directly within the ranking model's backbone. Unlike previous methods that applied multi-tasking only after the backbone, MORE uses Shared and Private Anchors to allow task-specific signals to evolve alongside representations at every layer. This approach has demonstrated superior performance on large-scale industrial datasets and, when deployed on Momo, a social discovery platform, resulted in significant improvements in user engagement metrics and a reduction in scoring latency. AI
IMPACT Enhances efficiency and effectiveness of ranking models in large-scale recommender systems.
RANK_REASON Publication of a new research paper detailing a novel AI architecture. [lever_c_demoted from research: ic=1 ai=1.0]
Read on arXiv cs.IR (Information Retrieval) →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
