Graph-Native Reinforcement Learning Enables Traceable Scientific Hypothesis Generation through Conceptual Recombination

• 类型：arxiv
• 标识：2607.00924
• 链接：https://arxiv.org/abs/2607.00924
• 主分类：engineering
• 形态：method
• 被引：2
• 被引来源：Semantic Scholar
• S2被引：2
• OpenAlex被引：0
• 影响力被引：1
• TLDR：Graph-PRefLexOR is developed, a family of graph-native reasoning models fine-tuned with Group Relative Policy Optimization to organize reasoning into explicit phases for mechanism exploration, graph construction, pattern extraction, and hypothesis synthesis, establishing graph-native reinforcement learning as a pathway toward interpretable AI systems for scientific hypothesis generation in materials design and other scientific applications.
• OpenAlex ID：W7167098711
• OpenAlex DOI：10.48550/arxiv.2607.00924
• DOI：10.48550/arxiv.2607.00924
• DOI来源：OpenAlex
• 开放获取：green
• 开放获取链接：https://doi.org/10.48550/arxiv.2607.00924
• OpenAlex更新：2026-07-19
• 待LLM分类：否
• 标题中文：图原生强化学习通过概念重组实现可追溯的科学假设生成
• TLDR中文：本文提出了 Graph-PRefLexOR，这是一族基于图结构的推理模型，使用 Group Relative Policy Optimization 进行微调，将推理过程组织为显式阶段，分别用于机理探索、图构建、模式提取和假设合成，确立了面向图结构的强化学习作为通向可解释 AI 系统的路径，可应用于材料设计及其他科学领域的科学假设生成。
• 来源文件：
• /inbox/tom/_candidates/2026-07-02-agent-rag-longcontext-candidates.json
• [S2 enrich]
• [OpenAlex backfill]