@inproceedings{xing-etal-2025-llmsr,
    title = "{LLMSR}@{XLLM}25: A Language Model-Based Pipeline for Structured Reasoning Data Construction",
    author = "Xing, Hongrui  and
      Liu, Xinzhang  and
      Jiang, Zhuo  and
      Yang, Zhihao  and
      Yao, Yitong  and
      Wang, Zihan  and
      Deng, Wenmin  and
      Wang, Chao  and
      Song, Shuangyong  and
      Yang, Wang  and
      He, Zhongjiang  and
      Li, Yongxiang",
    editor = "Fei, Hao  and
      Tu, Kewei  and
      Zhang, Yuhui  and
      Hu, Xiang  and
      Han, Wenjuan  and
      Jia, Zixia  and
      Zheng, Zilong  and
      Cao, Yixin  and
      Zhang, Meishan  and
      Lu, Wei  and
      Siddharth, N.  and
      {\O}vrelid, Lilja  and
      Xue, Nianwen  and
      Zhang, Yue",
    booktitle = "Proceedings of the 1st Joint Workshop on Large Language Models and Structure Modeling (XLLM 2025)",
    month = aug,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/landing_page/2025.xllm-1.31/",
    pages = "342--350",
    ISBN = "979-8-89176-286-2",
    abstract = "In this paper, we present a novel pipeline for the XLLM Shared Task-III: Large Language Model for Structural Reasoning (LLM-SR). Our pipeline addresses key challenges in automatic process-reward training data construction, such as high manual annotation costs, limited accuracy of large models in structured data processing, and dependency on auxiliary information for validation. To overcome these limitations, we first decompose the construction process into extraction and validation phases. Leveraging model-generated annotations, we produce pseudo-labeled data and iteratively refine model performance. Second, by analyzing structured data patterns, we encode structural constraints into a rule-based module and fine-tune the model with Gradient Reward Policy Optimization (GRPO), significantly improving structured data extraction success rates. Finally, we train the model to generate critical responses that assess evidence-conclusion relationships, thus enhancing validation reliability. Experimental results demonstrate that our pipeline outperforms models with an order of magnitude more parameters and achieves the first position on the task."
}