---
title: "多语言多智能体规划失败的可操作诊断：TART 方法"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.03735v1"
permalink: "https://feed.astralor.com/items/itm_30c565194a8d490d5dd4fead"
published: "2026-08-04T14:29:14.000Z"
score: 80
channel: "ai"
tags: ["GAIA","TART","多智能体系统","多语言"]
---

# 多语言多智能体规划失败的可操作诊断：TART 方法

## AI 点评

> 这篇论文的价值在于把多语言 Agent 的失败从模糊的“语言资源不足”落到了具体的规划-接地环节，而且给出了能直接用的补救手段。对做 Agent 基础设施或低资源语言产品的团队来说，TART 的思路值得立刻尝试。

## 速览

多语言多智能体系统在非英语场景下性能显著下降，但此前研究很少揭示用户请求转化为可执行计划时任务关键信息如何丢失。该论文将规划器视为请求到动作的接口，从真实失败执行中归纳出规划-接地失败分类法，并提出 TART，将分类法关键方面显式引入规划器及下游子智能体。在多个语言、三种 LLM 骨干、两个数据集和两种智能体配置下，TART 均持续提升性能，在多语言 GAIA 上将先进系统准确率平均提升 5.6 个百分点。

## 关键点

- 从真实失败执行中归纳出多语言多智能体规划-接地失败分类法，并发现低资源语言中此类失败占比更高。
- 提出 TART（Taxonomy-Guided Actionable Representation），将分类法关键信息显式注入规划器与子智能体。
- 在多种语言、三个 LLM 骨干、两个数据集和两种 Agent 配置下均稳定提升性能。
- 多语言 GAIA 上使 SOTA 系统准确率在 11 种语言上平均提高 5.6 个百分点。

## 全文

Multilingual multi-agent systems exhibit substantial degradation beyond English, yet prior work rarely identifies how task-critical information is lost when user requests are converted into executable plans. We study the planner in a multi-agent system as the request-to-action interface and derive an actionable taxonomy of planning-grounding failures from failed real-world task executions. LLM-based analysis shows that these failures constitute an increasing share of unsuccessful executions as language-resource availability declines, with the strongest effects in low-resource languages. To test whether the taxonomy supports mitigation, we introduce TART, Taxonomy-Guided Actionable Representation, that makes the taxonomy's key aspects explicit to the planner and downstream sub-agents. Across multiple languages, three LLM backbones, two datasets, and two agentic configurations, TART consistently improves performance. On multilingual GAIA, it raises a state-of-the-art system's accuracy by 5.6 percentage points averaged across eleven languages spanning low- to high-resource settings.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_30c565194a8d490d5dd4fead