LiuHongwei1992/520634PatentsDataset
收藏资源简介:
这是一个包含520,634项中国发明专利和实用新型专利的采样数据集,用于PatentConceptor系统——一个基于专利的跨领域设计概念生成工具。数据集覆盖了1985年至2024年的专利申请年份,通过分层随机采样方法从中国国家知识产权局(CNIPA)约3329万专利全集中提取,确保了在年份、国际专利分类(IPC)部分(A-H部)以及专利类型(发明与实用新型)之间的均衡覆盖。数据以CSV格式提供,包含专利名称、专利类型、摘要文本、国际专利分类号、申请人类型、申请年份、申请人城市、IPC主分类号、IPC部类信息和采样年份标签等字段。数据集支持直接通过Hugging Face的datasets库加载,也可用于本地部署。
A sampled dataset of 520,634 Chinese invention and utility model patents used for the PatentConceptor system — a patent-based cross-domain design concept generation tool. The dataset spans application years from 1985 to 2024 and is created via stratified random sampling from the full corpus of approximately 33.29 million patents from the China National Intellectual Property Administration (CNIPA), ensuring proportional coverage across years, all 8 IPC sections (A–H), and a balance between invention patents and utility models. It is provided in CSV format with fields including patent title, patent type, abstract text, IPC classification, applicant type, application year, applicant city, primary IPC, IPC section info, and sampling year label. The dataset can be loaded directly using Hugging Faces datasets library or used for local deployment.




