发表机构
Purdue University; Stevens Institute of Technology; Cornell University; University of Texas at El Paso; Johns Hopkins University（普渡大学; 史蒂文斯理工学院; 康奈尔大学; 德克萨斯大学埃尔帕索分校; 约翰斯·霍普金斯大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究将扎根理论引入智能体轨迹分析，提出AutoTraceGT流程，在六个语料库上可恢复多数人类标注失败模式，其编码本在失败预测任务中优于LLM基线，为智能体行为分析提供可扩展工具。
AI 中文摘要
理解智能体行为需要可扩展至数千条轨迹的方法，能在预构建分类器失效的长且常不熟悉的任务中发现新模式。我们提出将扎根理论引入智能体轨迹分析：这一源自社会科学、已有六十年历史的定性方法，具备原则性饱和标准与从数据到理论的可审计追踪。我们提出AutoTraceGT（基于扎根理论的自动化轨迹分析），首个在智能体轨迹上自动化实施扎根理论的多智能体流程。它迭代执行开放式、轴向及理论编码直至饱和，生成适配各任务的行为分类体系。在六个轨迹语料库上，AutoTraceGT生成的编码本可恢复人类标注分类体系中73%-91%的失败模式，并发现这些分类体系遗漏的额外模式。涌现的理论叙述与先前专家描述相符。作为演绎特征空间使用时，该编码本在下游失败预测任务中，性能优于零样本与少样本LLM基线。这些结果表明，扎根理论为研究智能体实际行为的ML研究者与智能体开发者提供了可扩展的分析工具。
英文摘要
Understanding agent behavior requires methods that scale to thousands of trajectories and surface new patterns in long, often unfamiliar tasks where pre-built classifiers fall short. We propose to bring grounded theory into agent trajectory analysis: a six-decade-old qualitative method from the social sciences, with a principled saturation criterion and an auditable trail from data to theory. We propose AutoTraceGT (Automated Trace analysis through Grounded Theory), the first multi-agent pipeline that automates grounded theory on agent trajectories. It iteratively performs open, axial, and theoretical coding until saturation, producing a behavioral taxonomy tailored to each task. Across six trajectory corpora, AutoTraceGT produces codebooks that recover 73-91 percent of the failure modes in human-annotated taxonomies and surface additional patterns that those taxonomies miss. The emergent theoretical narrative aligns with prior expert accounts. Used as a deductive feature space, the codebook outperforms zero-shot and few-shot LLM baselines on downstream failure prediction. These results suggest Grounded Theory offers a scalable analytic tool for ML researchers and agent developers studying what agents actually do.
Comments33 pages. Accepted to the Findings of the 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026)
