arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2607.02186cs.AI
收藏

UA-ChatDev: 不确定性感知的多智能体协作实现可靠软件开发

UA-ChatDev: Uncertainty-Aware Multi-Agent Collaboration for Reliable Software Development

Department of Electrical;Computer Engineering Prairie View A\&M University, Prairie View, TX 77446, USA Email

Temitayo Olamilekan Ogunsusi, Lijun Qian, Xishuang Dong

AI总结:

提出UA-ChatDev框架,通过不确定性量化与阶段感知阈值校准,在软件开发多智能体协作中抑制幻觉传播,提升代码执行可靠性。

摘要:

软件开发是一项复杂任务,需要不同角色的智能体协同合作。大型语言模型(LLMs)使得自主多智能体软件开发框架成为可能,这些框架利用基于角色的协作来自动化需求分析、编码、测试和优化。然而,现有方法通常假设中间智能体输出同样可靠,这使得它们容易受到幻觉传播的影响——即早期开发阶段产生的错误决策被传递给下游智能体,并对最终软件质量产生负面影响。为了解决这一挑战,我们提出了UA-ChatDev,一种不确定性感知的多智能体软件开发框架,它将不确定性量化集成到智能体交互中。该框架引入了一种基于token级对数概率的轻量级不确定性估计机制,以评估智能体响应的置信度,并采用阶段感知阈值校准,当不确定性超过可接受水平时,选择性触发基于检索的验证。在SRDD基准上的大量实验表明,UA-ChatDev在完整性、可执行性、一致性和整体质量指标上始终优于现有的单智能体和多智能体软件开发框架。进一步的消融研究和通信分析验证了不确定性感知交互增强了代码执行可靠性。

英文摘要:

Software development is a complex task that demands cooperation among agents with diverse roles. Large language models (LLMs) have enabled autonomous multi-agent software development frameworks that leverage role-based collaboration to automate requirements analysis, coding, testing, and refinement. However, existing approaches typically assume that intermediate agent outputs are equally reliable, leaving them vulnerable to hallucination propagation, where incorrect decisions generated in early development phases are transferred to downstream agents and negatively impact final software quality. To address this challenge, we propose UA-ChatDev, an uncertainty-aware multi-agent software development framework that integrates uncertainty quantification into agent interactions. It introduces a lightweight uncertainty estimation mechanism based on token-level log probabilities to assess the confidence of agent responses and employs phase-aware threshold calibration to selectively trigger retrieval-based verification when uncertainty exceeds acceptable levels. Extensive experiments on the SRDD benchmark demonstrate that UA-ChatDev consistently outperforms existing single-agent and multi-agent software development frameworks across completeness, executability, consistency, and overall quality metrics. Further ablation studies and communication analyses verify that uncertainty-aware interactions enhance code execution reliability.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑