基于专利时序数据预测技术机会的方法与实证研究  

Method and Empirical Investigation of Forecasting Technology Opportunity with Time-Series Patent Data

作　　者：李冬梅[1] 宋志红[2] 

机构地区：[1]山西大学数学科学学院,山西太原030006 [2]山西大学管理与决策研究所,山西太原030006

出　　处：《中国科技论坛》2015年第12期39-44,共6页Forum on Science and Technology in China

基　　金：教育部人文社会科学研究项目(09YJC630146)

摘　　要：本文提出一种基于专利时序数据预测技术机会的马尔科夫区制转换方法,并以1977年1月—2014年6月半导体行业专利时序数据为例进行实证分析。实证结果表明,半导体行业专利授权数量处于上升状态与下降状态的交替频率较高,且专利授权数量处于上升状态的持续期明显弱于专利授权数量处于下降状态的持续期:专利授权数量处于上升状态的平均持续期约为4.167个月,专利授权数量处于下降状态的平均持续期约为13.699个月。与以往研究相比,本文提出的马尔科夫区制转换(MRS)方法可以更准确地预测某个特定领域出现技术机会的时间窗口。The paper provides an Markov Regime-Switching model for forecasting technology opportunity with time-series patent data from January 1977 to June 2014 in semiconductor industry. The empirical results indicate that the transition between expansion state and contraction state of authorized patent quantity is relatively quick, and that the duration of an expansion state which authorized patent quantity is increasing significantly shorter than that of a contraction state in semiconductor industry. The average duration of an expansion state is about 4. 167 months, while the average duration of a contraction state is about 13. 699 months. Compared with previous studies, the Markov Regime-Switching （MRS） method in this paper provides a more accurate forecast on windows of technology opportunity in a specific field.

关 键 词：技术机会 马尔科夫区制转换 半导体行业 

分 类 号：G306.0[文化科学] C812[社会学—统计学]