AIGP: An LLM-Based Framework for Long-Term Value Alignment in E-Commerce Pricing
KDD 2026 ADS · Oral
An LLM-Agent framework that aligns short-term pricing with long-term platform and user value in e-commerce marketing.
About
I am a researcher at Alibaba Group, applying RL / Multi-Agent RL / Agentic RL to e-commerce marketing and pricing. My current focus is on RL-driven methods that advance LLM reasoning, planning, and long-horizon decision-making.
Previously at Xiamen University, I worked on cooperative MARL and neural activation dynamics in Deep RL.
Exploration > exploitation, in life too.
Research Interests
- Reinforcement Learning — policy optimization, value factorization, RL for LLMs
- Multi-Agent RL — cooperation, credit assignment, parameter sharing
- AI Agent & Agentic RL — planning under uncertainty, role-aware orchestration, long-horizon reasoning
- AI for E-Commerce — dynamic pricing, value alignment, marketing decision automation
News
- Jul 2026 Invited to serve as a Reviewer for AAAI 2027.
- Jun 2026 AIGP: An LLM-Based Framework for Long-Term Value Alignment in E-Commerce Pricing is selected as Oral at KDD 2026 ADS Track. 🎉
- Jun 2026 Invited to serve as a Reviewer for TMLR.
- May 2026 Recognized as a Gold Reviewer for ICML 2026.
- May 2026 AIGP: An LLM-Based Framework for Long-Term Value Alignment in E-Commerce Pricing accepted to KDD 2026 ADS Track.
- Mar 2026 Invited to serve as a Reviewer for NeurIPS 2026.
- Mar 2026 AIGP accepted to ICLR 2026 AIMS Workshop.
- Feb 2026 RASO: Role-Aware Shared Reflection for Multi-Agent Orchestration in E-Commerce Long-Horizon Planning accepted to ICAPS 2026.
- Feb 2026 Invited to serve as a Reviewer for ICLR 2026 AIMS Workshop.
- Jan 2026 SkillPrice: Semantic Skill Hierarchical Reinforcement Learning for Interpretable E-commerce Dynamic Price Recommendation accepted to DASFAA 2026.
- Dec 2025 Invited to serve as a Reviewer for ICML 2026.
- Sep 2025 PlanU: Large Language Model Decision Making through Planning under Uncertainty accepted to NeurIPS 2025.
- Sep 2025 Invited to serve as a Reviewer for ICLR 2026.
Selected Publications
Author names in bold indicate myself. Full list on Google Scholar.
AIGP: An LLM-Based Framework for Long-Term Value Alignment in E-Commerce Pricing
ICLR 2026 Workshop · AIMS
Workshop version presenting the long-term value alignment formulation for LLM-driven e-commerce pricing.
PlanU: Large Language Model Reasoning through Planning under Uncertainty
NeurIPS 2025
A planning-under-uncertainty framework that boosts LLM reasoning via explicit belief modeling and search over uncertain outcomes.
GradPS: Resolving Futile Neurons in Parameter Sharing Network for Multi-Agent Reinforcement Learning
ICML 2025
Diagnoses and repairs futile neurons in parameter-sharing MARL networks, improving cooperation and sample efficiency.
The Dormant Neuron Phenomenon in Multi-Agent Reinforcement Learning Value Factorization
NeurIPS 2024
Uncovers dormant neurons in MARL value factorization and proposes a reactivation strategy that restores network capacity and performance.
RiskQ: Risk-sensitive Multi-Agent Reinforcement Learning Value Factorization
NeurIPS 2023
A risk-sensitive value factorization framework for cooperative MARL, enabling reliable multi-agent decisions under uncertainty.
Experience & Education
-
2025 – Present
Researcher, Alibaba Group · Taobao & Tmall Group RL / MARL / Agentic RL for e-commerce marketing and pricing.
-
2022 – 2025
M.Eng. in Computer Science, Xiamen University · ASC Lab (spAtial Sensing and Computing) Advisor: Associate Prof. Siqi Shen. Cooperative MARL & neural activation dynamics in Deep RL.
-
2018 – 2022
B.Eng. in Software Engineering, Jilin University
Awards & Honors
- 2026 KDD 2026 ADS Track — Oral Presentation
- 2026 ICML 2026 Gold Reviewer
Academic Service
Reviewer for NeurIPS, ICML, ICLR, AAAI, and TMLR.
Contact
Open to research collaborations and discussions on RL, multi-agent systems, LLM agents, and AI for e-commerce.
✉ machennan.mcn@alibaba-inc.com
🏢 Alibaba Xixi Campus, Hangzhou