Meituan Shares 8 KDD 2026 Papers on Rec-Sys Models and Data Agents
美团技术团队 · wechat · 2026-08-13
Meituan's tech team shared 8 papers accepted by KDD 2026 and detailed their championship solution for the 2026 KDD Cup Data Agents track. Key highlights include:
- Rec-Sys & Training: MTFM, an alignment-free foundation model for industrial recommendation using heterogeneous tokens; MTGenRec, a PyTorch-based distributed training framework achieving 1.6x-2.4x speedup over TorchRec.
- Reward Modeling & Search: CDRRM, a contrast-driven rubric framework for reliable reward modeling with minimal samples; LocalSearchBench, a new benchmark evaluating agentic search in local life services.
- Ads & ETA: JTransNet and HMAF frameworks solving anonymity and multi-slot allocation in ad auctions; UME, a unified meta-generalization framework for zero-shot cross-domain ETA prediction.
- KDD Cup Solution: The team built a robust agent runtime supporting multimodal video understanding and document ETL, utilizing error feedback and auto-retries to ensure stability in long-chain data analysis tasks.
More from coding & agent
- Strong-to-Weak Test-Time Transfer: Harnesses Boost Weaker Models Without Parameter Updates — UIUC-CS · 2026-08-13
- Trajectory Labs Talk: The Distinction Between AI 'IQ' and 'Experience' — brianryhuang · 2026-08-13
- Google Proposes Agent Plugin Standard: Packaging Skills and MCP for Portability — Saboo_Shubham_ · 2026-08-13
- Developer Uses AI to Iteratively Build an Ant Mutation Simulator — breath_mirror · 2026-08-13
- Structured JSON Prompts for Video Generation: Porting Sora Prompts Directly to Minimax — ajrss2009 · 2026-08-13
- Anthropic Frontier Red Team Report: Multi-Agent Systems Prone to Echo Chambers and Consensus Herding — sebkrier · 2026-08-13