PatEx-Gym：面向专利审查新颖性评估的智能体现有技术检索框架
原标题：Novel or Not? PatEx-Gym: Agentic Prior-Art Search for Novelty Assessment in Patent Examination
AI 摘要
论文提出 PatEx-Gym，首个面向专利审查中新颖性评估的智能体搜索与推理框架，整合了 PatEx-Bench 基准（1K 真实审查任务、超 100 万专利与 5.2 万科学论文检索语料及支持迭代搜索的运行时）、无需人工标注的 PASS 轨迹合成方法（利用文档访问作为隐式相关性信号）以及基于合成轨迹监督微调的 4B 规模 PatEx-Agent。实验表明仅看最终决定准确率不足以反映审查质量，现有前沿模型在现有技术发现上仍很困难，而 4B 的 PatEx-Agent 可媲美更大的通用智能体。该工作将专利审查定位为智能体搜索与深度研究智能体的高难度测试平台。
正文节选
PatRe: A Full-Stage Office Action and Rebuttal Generation Benchmark for Patent Examination Novel or Not? PatEx-Gym: Agentic Prior-Art Search for Novelty Assessment in Patent Examination Abstract Patent examination is a challenging real-world task that requires agents to discover sparse prior art, iteratively refine queries, inspect technical evidence, and produce novelty assessments documented in Office Actions (OAs) across heterogeneous scientific and technical domains. However, existing benchm