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 2603.04384cs.CL
收藏

AgentIR: 为深度研究代理设计的推理感知检索

AgentIR: Reasoning-Aware Retrieval for Deep Research Agents

University of Waterloo(滑铁卢大学);University of Queensland(昆士兰大学);Carnegie Mellon University(卡内基梅隆大学)

Zijian Chen, Xueguang Ma, Shengyao Zhuang, Jimmy Lin, Akari Asai, Victor Zhong

AI总结:

AgentIR通过结合推理轨迹与查询的联合嵌入和数据合成方法,提升深度研究代理的检索性能,实现68%的准确率。

摘要:

深度研究代理正在迅速成为现代检索系统的主要消费者。与人类用户不同,人类用户在发出和细化查询时不会记录中间的思考过程,而深度研究代理在每次搜索调用前都会生成明确的自然语言推理,揭示了丰富的意图和上下文信息,而现有的检索器完全忽视了这些信息。为了利用这个被忽视的信号,我们引入了:(1) 推理感知检索,一种将代理的推理轨迹与其查询联合嵌入的检索范式;以及(2) DR-Synth,一种数据合成方法,它从标准问答数据集中生成深度研究检索器训练数据。我们证明了这两个组件都是独立有效的,它们的结合产生了一个训练好的嵌入模型AgentIR-4B,取得了显著的提升。在具有挑战性的BrowseComp-Plus基准测试中,AgentIR-4B在开放权重代理Tongyi-DeepResearch上实现了68%的准确率,相比传统嵌入模型两倍大小的模型50%的准确率,以及BM25的37%。代码和数据可在:https://texttron.github.io/AgentIR/上获得。

英文摘要:

Deep Research agents are rapidly emerging as primary consumers of modern retrieval systems. Unlike human users who issue and refine queries without documenting their intermediate thought processes, Deep Research agents generate explicit natural language reasoning before each search call, revealing rich intent and contextual information that existing retrievers entirely ignore. To exploit this overlooked signal, we introduce: (1) Reasoning-Aware Retrieval, a retrieval paradigm that jointly embeds the agent's reasoning trace alongside its query; and (2) DR-Synth, a data synthesis method that generates Deep Research retriever training data from standard QA datasets. We demonstrate that both components are independently effective, and their combination yields a trained embedding model, AgentIR-4B, with substantial gains. On the challenging BrowseComp-Plus benchmark, AgentIR-4B achieves 68\% accuracy with the open-weight agent Tongyi-DeepResearch, compared to 50\% with conventional embedding models twice its size, and 37\% with BM25. Code and data are available at: https://texttron.github.io/AgentIR/.

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。

↑