Finding Where the Buck Stops: An Automated Failure Attribution-Based Reflection Framework for Multi-Agent Collaboration · CiteArk

Finding Where the Buck Stops: An Automated Failure Attribution-Based Reflection Framework for Multi-Agent Collaboration

公开

作者：Xiaoqing Wang、Keman Huang、Bin Liang、Hongyu Li、Xiaoyong Du、Wuqiong Pan

EMNLP 2026 arXiv 2026 人工智能计算与语言机器学习 (stat.ML)

复现结果

全部结论 ↗

尚无复现结果

论文中的结论可在「研究结论」中查看。

完整研究主张目录

0 / 4 条结论已通过验证

其余仍在验证中

关键指标论文报告值 → 复现观测值 · 点击行展开详情

Across five trials, DoCtOR improves initial task success rates by 22% on HotPotQA, 26% on ChartQAPro, and 27% on Mind2Web, and outperforms the reported reflection baselines.整组结论：等待复现0/3 项指标获得支持 · 0 项已有评估3项测量结果展开结果表

实验方案已生成，还没有运行记录

测量结果

已评估 0/3 项

获得支持 0有争议或冲突 0尚未评估 3

保留同一结论下的完整结果组。已有评估排在前面；单项判断不替代整组结论。展开行查看条件、来源与历次证据。

每行是一项测量，分别显示论文报告、最近可用实测和累计证据状态。
指标与条件	论文报告	最近可用实测	证据判断
trials: 5 · dataset: HotPotQA · metric_scope: task success rate · test_tasks_per_trial: 100	22 percentage_points	—	尚未评估
trials: 5 · dataset: ChartQAPro · metric_scope: task success rate · test_tasks_per_trial: 100	26 percentage_points	—	尚未评估
trials: 5 · dataset: Mind2Web · metric_scope: task success rate · test_tasks_per_trial: 100	27 percentage_points	—	尚未评估

平台或执行故障 · 未形成科研结论

所需输入不可用 · 等待所需代码、数据或输入补齐后继续。

打开 Claim 详情

ProFA achieves the strongest reported agent-level and step-level failure-attribution accuracies among the listed methods on the held-in HotPotQA, ChartQAPro, and Mind2Web datasets and on the held-out Algorithm-Generated and Hand-Crafted subsets.整组结论：等待复现0/3 项指标获得支持 · 0 项已有评估3项测量结果展开结果表

实验方案已生成，还没有运行记录

测量结果

已评估 0/3 项

获得支持 0有争议或冲突 0尚未评估 3

保留同一结论下的完整结果组。已有评估排在前面；单项判断不替代整组结论。展开行查看条件、来源与历次证据。

每行是一项测量，分别显示论文报告、最近可用实测和累计证据状态。
指标与条件	论文报告	最近可用实测	证据判断
split: held-in · method: ProFA · dataset: HotPotQA · value_scope: test set of constructed Who & When Pro dataset · accuracy_level: agent	0.85 fraction	—	尚未评估
split: held-out Algorithm-Generated · method: ProFA · dataset: Who & When · failure_logs: 184 · accuracy_level: step	0.53 fraction	—	尚未评估
split: held-out Hand-Crafted · method: ProFA · dataset: Who & When · failure_logs: 184 · accuracy_level: step	0.2 fraction	—	尚未评估

平台或执行故障 · 未形成科研结论

实验执行出错 · 不会自动重试，需要人工决定后续处理。

打开 Claim 详情

定性结论

Using only reasoning steps after the decisive error step produces reflection quality comparable to using the complete failure trajectory, while diagnosis, correction, and PPO each contribute to performance.实验受阻，详见具体原因

The claim is supported by plotted curves without machine-readable scalar values, and direct evaluation requires the unavailable datasets, model assets, external agent services, and exact implementation details.

打开 Claim 详情

其余结论1

The evaluation is limited to English-language task-oriented collaborations with relatively small multi-agent teams and leaves open performance on open-ended or culturally diverse tasks and larger teams.论文自述边界

This is an explicit limitation note, not an independent empirical measurement requiring reproduction.

打开 Claim 详情