用树状推理和动作记忆，让AI网页代理更智能高效地完成复杂任务。
Branch-and-Browse: Efficient and Controllable Web Exploration with Tree-Structured Reasoning and Action Memory
- 采用树状结构分解任务，支持多分支推理与可控回溯。
- 在WebArena上达成35.8%成功率，执行时间减少40.4%。
- 适合需要精细控制与长期记忆的自动化网页操作场景。
基于大语言模型的自主网页代理在信息检索、报告生成和在线交易等目标导向任务中展现出巨大潜力，是实现开放网络环境中具身推理的关键一步。然而，现有方法在推理深度和效率上仍受限：单纯的线性方法难以处理多步推理且缺乏有效回溯能力，而其他搜索策略则粒度粗、计算成本高。我们提出Branch-and-Browse，一种细粒度网页代理框架，统一了结构化推理-执行、上下文记忆与高效执行机制。该框架（i）通过树状结构探索实现显式的子任务管理，支持可控的多分支推理；（ii）通过后台推理实现网页状态高效重播，加速探索过程；（iii）利用页面动作记忆，在会话内及跨会话间共享已探索动作。在WebArena基准测试中，Branch-and-Browse达到35.8%的任务成功率，并相较最先进方法将执行时间减少40.4%。结果表明，Branch-and-Browse是一种可靠且高效的基于LLM的网页代理框架。
原文摘要 · Abstract (English)
Autonomous web agents powered by large language models (LLMs) show strong potential for performing goal-oriented tasks such as information retrieval, report generation, and online transactions. These agents mark a key step toward practical embodied reasoning in open web environments. However, existing approaches remain limited in reasoning depth and efficiency: vanilla linear methods fail at multi-step reasoning and lack effective backtracking, while other search strategies are coarse-grained and computationally costly. We introduce Branch-and-Browse, a fine-grained web agent framework that unifies structured reasoning-acting, contextual memory, and efficient execution. It (i) employs explicit subtask management with tree-structured exploration for controllable multi-branch reasoning, (ii) bootstraps exploration through efficient web state replay with background reasoning, and (iii) leverages a page action memory to share explored actions within and across sessions. On the WebArena benchmark, Branch-and-Browse achieves a task success rate of 35.8\% and reduces execution time by up to 40.4\% relative to state-of-the-art methods. These results demonstrate that Branch-and-Browse is a reliable and efficient framework for LLM-based web agents.
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。