arXiv cs.AI· M. Yigit Turali, Shreyas Rajesh, M. Efe Lorasdagi, Chenda Duan, Kartik Sharma, Vwani Roychowdhury·· 3 天前AI 评分36
PANINI-DR：通过生成式语义工作空间实现可控深度研究
PANINI-DR: Controlled Deep Research through Generative Semantic Workspaces
AI 导读
PANINI-DR 是一个多智能体深度研究框架，将整体调查协调与子问题研究、验证分离，基于 Generative Semantic Workspace 构建共享工作空间记录子问题、依赖关系、发现与证据。
正文 · 原文
Abstract:Deep research requires connecting multiple steps of search, reasoning, and computation, with intermediate findings guiding what to investigate next. As investigations grow, agents must maintain research state across model calls and check findings before errors propagate into subsequent work. We introduce PANINI-DR, a multi-agent harness that separates coordination of the overall investigation from the research and verification of individual subquestions. A shared workspace, built on the Generative Semantic Workspace representation, records subquestions, their dependencies, findings, and supporting evidence. Researchers investigate individual subquestions, while a separate verification agent checks proposed answers against those subquestions and their evidence. Verification feedback guides further investigation, and recorded findings inform subsequent assignments. We evaluate PANINI-DR on BrowseComp-Plus, FRAMES, and MoNaCo against seven state-of-the-art deep-research harnesses, all using a shared language-model backbone and frozen corpora. PANINI-DR achieves the strongest average performance across the three benchmarks. On BrowseComp-Plus, it achieves 73.3% accuracy, outperforming the next-best harness by 16.0 points. Its advantage is strongest on hard questions, where it answers nearly twice as many as the next-best harness. Ablations further show that structured planning and persistent state improve performance, as does checking each finding against its cited evidence before it is reused. Because a model can answer correctly from memory alone, we also analyze the evidence retained by each harness. PANINI-DR achieves the highest answer recoverability from saved evidence on all three benchmarks, while complementary claim-support audits show stronger evidence-backed answers. Accounting for this evidence changes how harnesses compare and further strengthens PANINI-DR's advantage.
| Comments: | Under review at ICLR 2027. 37 pages, 1 figure, 22 tables. M. Yigit Turali and Shreyas Rajesh contributed equally |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR) |
| ACM classes: | I.2.7; I.2.11; H.3.3 |
| Cite as: | arXiv:2610.15208 [cs.CL] |
| (or arXiv:2610.15208v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.15208 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Mehmet Yigit Turali [view email]
[v1]
Sat, 10 Oct 2026 22:42:08 UTC (583 KB)
来源：arXiv cs.AI · arxiv.org