今日速览
| 序号 | 标题 | 来源 | 日期 | 主题 | 推荐等级 |
|---|---|---|---|---|---|
| 1 | OpenRCA 2.0: From Outcome Labels to Causal Process Supervision | arXiv | 2026-06-25 | RAG | 高 |
| 2 | Joint Learning of Experiential Rules and Policies for Large Language Model Agents | arXiv | 2026-06-25 | AI-Agent | 高 |
重点论文与技术动态
1. OpenRCA 2.0: From Outcome Labels to Causal Process Supervision
- 来源：arXiv
- 日期：2026-06-25
- 作者/机构：Aoyang Fang, Yifan Yang, Jin’ao Shang, Qisheng Lu, Junjielung Xu, Rui Wang
- 主题标签：
RAG,arXiv - 推荐等级：高
- 分类：cs.AI
一句话结论
OpenRCA 2.0通过步骤式因果标注揭示LLM在根因分析中的核心缺陷，仅20.7%案例能精确恢复根因集。
核心内容
- 现有RCA数据集仅标注根因，缺失因果传播路径，使任务简化为模式匹配。
- PAVE协议利用故障注入重建因果路径，采用前向验证机制（从因到果）而非反向推理。
- OpenRCA 2.0基准测试显示，LLMs根因识别率达76.0%，但因果路径验证率仅61.5%，暴露“未验证诊断”失败模式。
方法与数据
- PAVE协议基于故障注入干预重建因果传播路径。
- OpenRCA 2.0含500个实例，是首个跨系统步骤式因果标注RCA基准。
价值判断
- 值得关注：步骤式因果标注能暴露“未验证诊断”失败模式，现有仅结果评估掩盖此问题。
- 可复用点：PAVE协议可推广至其他因果推理任务的基准构建。
- 局限/待核查：LLMs在因果路径验证成功率低（61.5%），需提升模型推理能力。
摘要
Root cause analysis (RCA) poses a holistic test of LLM agentic capabilities, such as long-context understanding, multi-step reasoning, and tool use. However, existing datasets suffer from a fundamental gap: they label only the root cause, not the propagation path connecting it to the observed symptom, which largely simplifies the task to naive pattern matching. To support rigorous evaluation, we introduce PAVE, a step-wise labeling protocol that leverages known interventions from fault injection to reconstruct causal propagation paths. The mechanism is forward verification: reasoning from cause to effect rather than inferring backward from symptoms. Applying PAVE yields OpenRCA 2.0 (500 instances), the first cross-system RCA benchmark with step-wise causal annotations for LLM agents. Across 11 frontier LLMs, recovering the exact root-cause set succeeds in only 20.7% of cases on average. To locate where this difficulty lies, we relax the criterion and find what we call the ungrounded diagnosis: agents identify at least one correct root-cause service in 76.0% of cases, but ground that service in a verified causal propagation path to the observed symptom in only 61.5%. Outcome-only evaluation hides this failure mode; step-wise causal ground truth is the missing piece for trustworthy LLM-based RCA agents.2. Joint Learning of Experiential Rules and Policies for Large Language Model Agents
- 来源：arXiv
- 日期：2026-06-25
- 作者/机构：Shicheng Ye, Chao Yu
- 主题标签：
AI-Agent,arXiv - 推荐等级：高
- 分类：cs.AI
一句话结论
JERP通过联合学习经验规则和策略，显著提升了LLM代理在复杂交互任务中的决策性能。
核心内容
- 现有方法分离经验使用：规则外部化易解释但易脱节，轨迹更新广义改进但局部修正有限（稀疏奖励设置下）；JERP耦合规则池与策略更新。
- 决策时检索任务相关规则并结合交互历史条件化代理。
- 每个回合后，利用收集轨迹优化策略并修订规则池，通过比较当前轨迹与参考成功轨迹。
方法与数据
- 方法：基于交互轨迹联合优化规则池和策略。
- 数据：在AlfWorld和WebShop数据集上验证。
价值判断
- 值得关注：有效解决经验规则与策略脱节问题，提升决策稳定性。
- 可复用点：联合学习框架可迁移至其他多步交互式LLM任务。
- 局限/待核查：摘要未明确提及局限。