---
格式版本: 2
标题: "Dude: A Dual-Detection Multi-Agent System for Paper-Code Discrepancy Detection"
原文链接: "https://arxiv.org/abs/2609.03416"
发布日期: "2026-09-03"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Thu, 3 Sep 2026 06:24:33 UTC (939 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-09-06T21:26:56+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-09-06T21:24:50+08:00"
入库时间: "2026-09-06T13:26:56.316Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=Scale-up&searchtype=all"
匹配关键词:
  - "Scale-up"
  - "performance"
  - "AI"
相关厂家:
  []
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 20
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "内容为arXiv论文，主题是论文代码差异检测的多智能体系统，与超节点/AI Rack/机柜级AI基础设施完全无关，仅搜索词命中Scale-up，实质无任何相关技术或商业信息。"
AI质检模型: "zj-deepseek-v4-flash"
AI质检时间: "2026-09-06T21:27:04+08:00"
AI主题相关性: 0
AI来源权威性: 10
AI新颖性: 5
AI技术细节: 0
AI商业部署信号: 0
AI完整性: 5
AI摘要: "本文提出Dude，首个用于论文与代码差异检测的双检测多智能体系统，以解决现有单智能体方法召回率低的问题。该系统引入粒度对齐协商与两阶段显著性过滤机制，在真实数据集上相比基线方法召回率最高提升22.8%，F1分数最高提升18.7%。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-09-06T23:44:05.000Z"
采集批次: "2026年9月6日19点27分09秒"
采集批次ID: "20260906-192709-237"
去重键: "https://arxiv.org/abs/2609.03416"
---

## Computer Science > Artificial Intelligence

## Title:Dude: A Dual-Detection Multi-Agent System for Paper-Code Discrepancy Detection

[View PDF](https://arxiv.org/pdf/2609.03416) [HTML (experimental)](https://arxiv.org/html/2609.03416v1)

> Abstract:LLM-empowered paper-code discrepancy detection has received growing concern since the scaling of research submissions exceeds the manual review capability. However, the limited context capacity and one-sided discrepancy detection of existing single-agent LLM paradigms lead to an inferior recall performance in detecting discrepancies. In this paper, we propose Dude, the first Dual-Detection Multi-Agent System for paper-code discrepancy detection. We discover that the granularity asymmetry of the paper-language and code-language introduces over-interpretation and over-reporting challenges in a multi-agent system design for discrepancy detection, resulting in increasing false positives. To address this, we propose a granularity-aligned negotiation and a two-stage salience-filtering mechanism in Dude, which effectively prevents agents from falsely reporting discrepancies. Experimental results in real-world paper-code discrepancy datasets showcase Dude's significant recall and precision improvement by up to 22.8%, increasing F1 score by up to 18.7% compared to baseline methods.

| Comments: |  |
| --- | --- |
| Subjects: | Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | [arXiv:2609.03416](https://arxiv.org/abs/2609.03416) \[cs.AI\] |
|  | (or [arXiv:2609.03416v1](https://arxiv.org/abs/2609.03416v1) \[cs.AI\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2609.03416](https://doi.org/10.48550/arXiv.2609.03416) |

## Submission history

From: Weijie Liu \[[view email](https://arxiv.org/show-email/23f3e1d2/2609.03416)\]  
**\[v1\]** Thu, 3 Sep 2026 06:24:33 UTC (939 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2609.03416) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))