发表机构
Monash University; The University of Queensland; Xiamen University（莫纳什大学; 昆士兰大学; 厦门大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出SWORD框架，联合优化多代理工作流与提示，仅凭标量指标驱动，超越多种基线，以更小模型和低成本提升精度，并自主发现领域特征。
AI 中文摘要
用户行为模拟是通过使用模拟代理代替真实用户，对信息系统中用户交互进行计算建模的过程。它支持系统测试与评估、决策与预测以及用户体验设计。现有模拟器依赖手工规则或领域专业知识，这些知识在不同任务间迁移效果不佳。本文提出了SWORD（基于角色设计的模拟驱动工作流与提示优化）框架，该框架联合优化多代理工作流拓扑结构和自然语言提示。它仅由标量任务指标引导，无需领域初始化或特定任务的工程干预。实验结果表明，在受控的相同骨干模型比较下，SWORD相比仅提示优化、仅工作流优化和分阶段优化基线取得了统计学上显著的性能提升。与已发表的最强领域特定基线相比，SWORD在使用更小骨干模型、显著更少训练数据和非常合理的API成本（每个数据集4至6美元）的情况下，进一步提高了准确性。除了预测性能，SWORD还能仅从标量误差反馈中自主发现领域相关信号、评论情感映射规则和流行病学衰减先验，从而建立文本梯度作为用户行为建模中无监督特征重要性发现的机制。
英文摘要
User behavior simulation is the computational modeling of user interactions within information systems through the use of simulated agents in place of live users. It supports system testing and evaluation, decision-making and forecasting, and user experience design. Existing simulators rely on hand-crafted rules or domain expertise that transfers poorly across tasks. SWORD (Simulation-driven Workflow and Prompt Optimization with Role-based Design) is introduced as a framework that jointly optimizes multi-agent workflow topology and natural-language prompts. It is guided solely by a scalar task metric, without domain initialization or task-specific engineering. The experimental results demonstrate that SWORD achieves statistically significant gains over prompt-only, workflow-only, and staged-optimization baselines under a controlled, identical-backbone comparison. Against the strongest published domain-specific baseline, SWORD further improves accuracy while using a smaller backbone model, substantially less training data, and a very reasonable API cost (\$4--\$6 for each dataset). Beyond predictive performance, SWORD autonomously discovers domain-relevant signals, review-sentiment mapping rules and epidemiological decay priors, purely from scalar error feedback, establishing textual gradients as a mechanism for unsupervised feature-importance discovery in user behavior modeling.
Commentsunder review for ACM Transactions on Information Systems Journal, 34 pages, 2 figures
