arXiv：cs.AI· Ziyu Chen, Yilun Zhao, Jiashuo Sun, Yiling Ma, Manasi Patwardhan, Arman Cohan·· 6 小时前AI 评分33
IdeaAnchor：训练 LLM 将文献转化为研究想法
IdeaAnchor: Teaching LLMs to Turn Literature into Research Ideas
AI 导读
IdeaAnchor 提出一种用结构化规范作为特权信号训练 LLM 做研究构思的新范式，每个实例编码输入论文的功能角色、关系与目标综合标准，数据从已发表论文中挖掘。团队通过示范、自蒸馏和强化学习训练模型，并在推理时加入检索增强。实验显示构思质量持续提升：anchor 训练强化创造性综合，检索改善细节展开，两者结合效果最佳。
正文
Abstract:Scientific research often begins by synthesizing ideas from a set of related papers to identify gaps and formulate new directions. However, training language models to perform this form of literature-grounded ideation remains challenging, as existing approaches based on prompting or feedback lack structured supervision for how papers should be synthesized. We introduce IdeaAnchor, a paradigm for training LLMs to perform research ideation using structured specifications as privileged signals. Each IdeaAnchor instance encodes how each input paper should be synthesized into a successful idea, including their functional roles, relationships, and target synthesis criteria. We build this paradigm by mining instances from published papers, capturing how real ideas emerge from prior literature. We then train models via demonstration, self-distillation, and reinforcement learning, and further enhance generation with retrieval at inference time. Experiments show consistent improvements in ideation quality. Our analysis reveals a functional decomposition: anchor-based training strengthens creative synthesis, retrieval enhances detail elaboration, and combining both yields the best performance.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.08781 [cs.CL] |
| (or arXiv:2610.08781v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.08781 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Ziyu Chen [view email]
[v1]
Tue, 6 Oct 2026 17:59:01 UTC (2,683 KB)
来源：arXiv：cs.AI · arxiv.org