专利转化特征精准识别与预测——以人工智能芯片为例  

Accurate Identification and Prediction of Patent Transformation Features:An Example of Artificial Intelligence Chip

作　　者：姜南 李逸凡 刘谦[2] 刘星 Jiang Nan;Li Yifan;Liu Qian;Liu Xing(Shanghai International College of Intellectual Property, Tongji University, Shanghai 200092, China;Business School of Hohai University,Nanjing 210024, China)

机构地区：[1]同济大学上海国际知识产权学院,上海200092 [2]河海大学商学院,江苏南京210024

出　　处：《科技进步与对策》2022年第10期1-10,共10页Science & Technology Progress and Policy

基　　金：国家社会科学基金重大项目(17ZDA140);国家自然科学基金项目(71874122);同济大学研究生课程思政研究项目(2021KCSZYJ28)。

摘　　要：探讨前沿科技领域专利转化特征并对其进行精准识别与预测,对于我国破解“卡脖子”技术难题及实现科技自立自强具有重要意义。选取人工智能芯片专利领域,采用机器学习算法测度最优转化预测方案,分析全球范围内主要国家或地区专利成功转化影响因素,从企业/高校、国内/国际等不同层面总结专利成功转化的主要特征。结果发现:随机森林算法预测效果较好,人工智能芯片领域专利转化概率服从对数曲线分布,影响高校/企业、国内/国外专利转化特征的因素有所不同。最后,提出高校/科研机构应注重高价值专利维持和团队合作、企业应提升专利技术质量和撰写质量等政策建议。Although there has been an exponential growth of patent applications in China,the commercialization potential of the majority of patents registered in the field of science and technology has not been exploited.Based on this background,the paper studies the identification factors of patent transformation in order to better promote patent transformation.The commercialization value of patents mainly depends on patent protection and patent policy.Given the current government policies on patents,more data are needed to predict patent technology transformation in emerging frontier fields,help to solve the problems that China has to rely on foreign key core technologies for the production of some high-tech products and realize the commercialization potential of science and technology patents in China.The paper aims to better promote the patent transformation of key core technologies and focuses on artificial intelligence(AI)chips,given their current importance.In this study,AI chips patent data is from the Derwent World Patents Index Database.We employ logical regression,support vector machine,random forest and AdaBoost algorithms to carry out comparative method analysis.After decomposing the patent transformation indicators into three dimensions(technology,law and economy),we select 17 representative indicators in the field of AI chips and adopt the machine learning method to identify an optimal transformation prediction scheme and the factors influencing successful patent transformation in China and in other countries.We discuss the main characteristics of successful patent transformation in different areas and fields with various application bodies in China and abroad.This study applies different algorithms to analyse patent transformation in the field of AI chips in China and a number of developed countries.These algorithms help to predict the main factors influencing the successful transformation of patent technology in the field of AI chips.Among the four algorithms(logical regression,support vector machine,random

关 键 词：专利转化 机器学习 随机森林算法 科技成果转化 人工智能芯片 

分 类 号：G306[文化科学]