@inproceedings{hu-etal-2025-webcot,
    title = "{W}eb{C}o{T}: Enhancing Web Agent Reasoning by Reconstructing Chain-of-Thought in Reflection, Branching, and Rollback",
    author = "Hu, Minda  and
      Fang, Tianqing  and
      Zhang, Jianshu  and
      Ma, Jun-Yu  and
      Zhang, Zhisong  and
      Zhou, Jingyan  and
      Zhang, Hongming  and
      Mi, Haitao  and
      Yu, Dong  and
      King, Irwin",
    editor = "Christodoulopoulos, Christos  and
      Chakraborty, Tanmoy  and
      Rose, Carolyn  and
      Peng, Violet",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
    month = nov,
    year = "2025",
    address = "Suzhou, China",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/author-page-yu-wang-polytechnic/2025.findings-emnlp.276/",
    doi = "10.18653/v1/2025.findings-emnlp.276",
    pages = "5155--5173",
    ISBN = "979-8-89176-335-7",
    abstract = "Web agents powered by Large Language Models (LLMs) show promise for next-generation AI, but their limited reasoning in uncertain, dynamic web environments hinders robust deployment. In this paper, we identify key reasoning skills essential for effective web agents, i.e., reflection {\&} lookahead, branching, and rollback, and curate trajectory data that exemplifies these abilities by reconstructing the agent{'}s (inference-time) reasoning algorithms into chain-of-thought rationales. We conduct experiments in the agent self-improving benchmark, OpenWebVoyager, and demonstrate that distilling salient reasoning patterns into the backbone LLM via simple fine-tuning can substantially enhance its performance. Our approach yields significant improvements across multiple benchmarks, including WebVoyager, Mind2web-live, and SimpleQA (web search), highlighting the potential of targeted reasoning skill enhancement for web agents."
}