Mango通过全局视角选择网页导航起点，提升复杂网站探索效率。
Mango: Multi-Agent Web Navigation via Global-View Optimization
- 用多臂赌博机模型动态选起点，自适应分配探索预算
- 在WebVoyager上达63.6%成功率，比最佳基线高7.3%
- 支持开源与闭源模型，适合需高效网页导航的研究者
现有网页智能体通常从根网址开始探索，面对结构复杂的深层网站效率低下。缺乏全局网站结构视图导致智能体易陷入导航陷阱、误入无关分支或在有限预算内无法抵达目标信息。我们提出Mango，一种基于网站结构动态确定最优起始点的多智能体网页导航方法。将URL选择建模为多臂赌博机问题，采用Thompson Sampling自适应分配导航预算。此外，引入情景记忆组件存储导航历史，使智能体可从过往尝试中学习。在WebVoyager上的实验表明，使用GPT-5-mini时Mango的成功率达63.6%，优于最佳基线7.3%。在WebWalkerQA上，成功率达到52.5%，超越最佳基线26.8%。我们还验证了Mango在开源与闭源模型上的泛化能力。数据与代码已开源，详见https://github.com/VichyTong/Mango。
原文摘要 · Abstract (English)
Existing web agents typically initiate exploration from the root URL, which is inefficient for complex websites with deep hierarchical structures. Without a global view of the website's structure, agents frequently fall into navigation traps, explore irrelevant branches, or fail to reach target information within a limited budget. We propose Mango, a multi-agent web navigation method that leverages the website structure to dynamically determine optimal starting points. We formulate URL selection as a multi-armed bandit problem and employ Thompson Sampling to adaptively allocate the navigation budget across candidate URLs. Furthermore, we introduce an episodic memory component to store navigation history, enabling the agent to learn from previous attempts. Experiments on WebVoyager demonstrate that Mango achieves a success rate of 63.6% when using GPT-5-mini, outperforming the best baseline by 7.3%. Furthermore, on WebWalkerQA, Mango attains a 52.5% success rate, surpassing the best baseline by 26.8%. We also demonstrate the generalizability of Mango using both open-source and closed-source models as backbones. Our data and code are open-source and available at https://github.com/VichyTong/Mango.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。