@inproceedings{chang-etal-2026-enhancing,
    title = "Enhancing Agentic Textual Graph Retrieval with Synthetic Stepwise Supervision",
    author = "Chang, Ge  and
      Su, Jinbo  and
      Liu, Jiacheng  and
      Yang, Pengfei  and
      Shang, Yuhao  and
      Zheng, Huiwen  and
      Ma, Hongli  and
      Liang, Yan  and
      Li, Yuanchun  and
      Liu, Yunxin",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.acl-long.1169/",
    pages = "25510--25525",
    ISBN = "979-8-89176-390-6",
    abstract = "Integrating textual graphs into Large Language Models (LLMs) is promising for complex graph-based QA. However, a key bottleneck is retrieving informative yet compact subgraphs that fit the LLM context. Existing retrievers often struggle, relying either on shallow embedding similarity or costly interactive policies that require excessive supervision.To address these challenges, we introduce Graph-S$^3$, an agentic textual graph reasoning framework featuring an LLM-based retriever trained with synthetic stepwise supervision. Rather than relying on final answer rewards{---}which often yield sparse and unstable signals{---}we optimize the retriever by evaluating each step against offline-extracted golden subgraphs.Our approach distills golden subgraphs via a specialized data synthesis pipeline to formulate dense rewards, facilitating a two-stage training scheme that effectively learns the interactive graph exploration policy.Based on extensive experiments on three common datasets in comparison with seven strong baselines, our approach achieves an average improvement of 15.6{\%} in accuracy and 17.2{\%} in F$_1$ score. The advantage is even higher in more complicated multi-hop reasoning tasks."
}