English(EN) Dude: A Dual-Detection Multi-Agent System for Paper-Code Discrepancy Detection

新的多智能体系统解决论文-代码差异问题

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-04 04:00

研究人员推出Dude，一种新颖的双重检测多智能体系统，旨在识别研究论文与其关联代码之间的差异。该系统解决了单智能体LLM方法的局限性，这些方法在上下文容量和单方面检测方面存在问题，导致召回率较低。Dude采用粒度对齐的协商和两阶段显著性过滤机制，以减轻由语言粒度差异引起的误报。实验表明，Dude显著提高了召回率和精确率，与现有方法相比，F1分数最高可提高18.7%。 AI

影响该系统可以通过更好地验证代码与其配套论文的一致性来提高研究可复现性的可靠性。

排序理由该条目是一篇学术论文，详细介绍了一种用于论文-代码差异检测的新系统。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 17 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该条目是一篇学术论文，详细介绍了一种用于论文-代码差异检测的新系统。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, model release; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Weijie Liu, Running Zhao, Wenhao Yuan, Jinfeng Xu, Zhanfeng Xu, Xiaoxi Zhang, Edith Cheuk-Han Ngai · 2026-09-04 04:00

Dude：一个用于论文-代码差异检测的双重检测多智能体系统

arXiv:2609.03416v1 Announce Type: new 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…

本文如何被排名

报道来源 [1]

Dude：一个用于论文-代码差异检测的双重检测多智能体系统

相关实体

相关话题