UserArena：面向长程购物智能体的交互式用户模拟基准
原标题：Paper page - UserArena: Benchmarking Interactive User Simulation for Long-Horizon Shopping Agents
AI 摘要
研究团队发布 UserArena，一个基于真实电商数据的交互式基准，包含来自 11,010 名用户的 64,729 条会话轨迹，覆盖 566,102 个商品和 13 种动作类型，并支持与生产 A/B 测试结果进行离线验证。该基准提出五级评估体系，从输出有效性、局部决策延伸到序列行为、个性化与最终结果。团队同时提出 On-Policy Trajectory Distillation（OPTD），利用轨迹级意图在学生生成的 rollout 上提供密集教师目标，实验显示其在评估协议下保真度最强，并最接近复现生产策略效果。
正文节选
UserArena: Benchmarking Interactive User Simulation for Long-Horizon Shopping Agents Abstract UserArena introduces a real-world interactive benchmark with long-horizon shopping trajectories and an evaluation framework, alongside On-Policy Trajectory Distillation to improve simulator fidelity for agentic recommendations. Agentic recommendation pursues shopper goals through sequences of context-dependent decisions. Developing and evaluating such policies requires faithful recommendation world mode