基于马尔可夫链的专利产业化概率模型:专利引证的视角

科研管理 ›› 2015, Vol. 36 ›› Issue (6): 10-19.

基于马尔可夫链的专利产业化概率模型:专利引证的视角

许琦^1,2, 顾新建¹

1. 浙江大学机械工程学系工业工程中心, 浙江杭州 310027;
2. 台州职业技术学院台州经济研究所, 浙江台州 318000

收稿日期:2014-01-15 修回日期:2014-10-30 出版日期:2015-06-25 发布日期:2015-06-23
通讯作者: 许琦
作者简介:许琦(1983-),男,浙江台州人,浙江大学机械工程学系博士生,研究方向:专利挖掘、知识网络;
顾新建(1956-),男,浙江杭州人,浙江大学机械工程学系教授,博士生导师,研究方向:知识管理,制造业信息化。
基金资助:
国家自然科学基金项目:"智慧的低碳设计和制造一体化技术及系统研究"(51175463, 2012.01-2015.12), 浙江省哲学社会科学规划课题"基于专利引证网络的知识基因提取方法探索"(13NDJC19YBM, 2013.08-2015.07), 浙江省软科学研究计划项目"技术标准下提升企业自主创新能力——基于专利池的组建与管理"(2013C35064, 2013.05-2015.08), 浙江省软科学研究计划项目"提升地方应用型高校专利产业化实效性研究——基于浙江省高校与企业的调研"(2014C35041, 2014.07-2016.06), 台州市科协软科学研究课题"提升企业核心竞争力路径研究:专利产业化判别与评价"(台科协[2014]43号, 2014.09-2015.01)

A patent industrialization probability model based on Markov Chain: The perspective of patent citation

Xu Qi^1,2, Gu Xinjian¹

1. Industrial Engineering Center, Department of Mechanical Engineering, Zhejiang University, Hangzhou 310027, Zhejiang, China;
2. Taizhou Economic Institute, Taizhou Vocational and Technical College, Taizhou 318000, Zhejiang, China

Received:2014-01-15 Revised:2014-10-30 Online:2015-06-25 Published:2015-06-23

摘要/Abstract

摘要： 从专利引证的视角, 运用数理统计的方法衡量专利在技术发展过程中的重要性和价值, 从定量分析入手对专利产业化的概率进行辨析。论证了专利产业化的判别与评价原理。专利引证网络的时间无关性、不可约性以及引证路径单向性满足马尔可夫链收敛的条件。运用马尔可夫链对专利引证网络的随机浏览过程进行建模。建立了专利引证网络与马尔可夫链的对应关系, 将专利产业化概率的计算过程转化为计算马尔可夫链转移矩阵稳态分布的过程, 设计了转移矩阵最大特征值和特征向量的迭代算法。从美国专利商标局的专利数据库中采集了235项1976年至2006年授权的半导体制造领域电子发射器制造技术的相关专利, 分析了专利的年度分布情况。从美国国家经济研究局的专利数据集中查询得到592条专利引证关系, 利用网络分析软件Pajek构建专利引证网络。实验结果表明, 本文提出专利产业化概率模型切实有效, 与专利产业化的实际情况相符, 可以作为专利产业化可能性的一种指示。

关键词: 专利产业化, 马尔可夫链, 专利引证网络, 转移矩阵, 电子发射器制造

Abstract: The article differentiates the patent industrialization probability with a quantitative analysis from the perspective of patent citation. The method of mathematical statistics is used to measure the importance and values of patents in the technology development process. The principle of identification and evaluation of patent industrialization is demonstrated. Time-independence, irreducibility and irreversibility of a patent citation network enable it to meet the conditions for convergence of Markov Chain. So Markov Chain is adopted to model the process of random browse in a patent citation network. The correspondence between a patent citation network and Markov Chain is established. And the calculation of patent industrialization probability is translated to calculate the stable distribution of the transfer matrix of Markov Chain. The iterative algorithm for computing the largest eigenvalues and eigenvectors of the transfer matrix is devised. The article collects 235 patents related to electron emitter manufacturing which were granted between 1976 and 2006 from the database of United States Patent and Trademark Office. The annual distribution of the patents is analyzed. The article queries from the patent data set of National Bureau of Economic Research to get 592 patent citations and uses the network analysis software Pajek to build a patent citation network as the experimental data. The experimental results show that, the proposed model is effective. It is consistent with the actual situations of patent industrialization, and can be used as the instructions of patent industrialization.

Key words: patent industrialization, Markov Chain, patent citation network, transfer matrix, electron emitter manufacturing

中图分类号:

G306.0

许琦, 顾新建. 基于马尔可夫链的专利产业化概率模型:专利引证的视角[J]. 科研管理, 2015, 36(6): 10-19.

Xu Qi, Gu Xinjian. A patent industrialization probability model based on Markov Chain: The perspective of patent citation[J]. Science Research Management, 2015, 36(6): 10-19.

参考文献

[1] 中华人民共和国国家统计局. 2012年国民经济和社会发展统计公报[EB/OL]. (2013-02-22)[2014-01-07].http://www.ce.cn/xwzx/gnsz/gdxw/201302/22/t20130222_24134444.shtml.National bureau of statistics of the people's republic of China. Statistical bulletin of national economic and social development 2012 [EB/OL]. (2013-02-22)[2014-01-07]. http://www.ce.cn/xwzx/gnsz/gdxw/201302/22/t20130222_24134444.shtml. [2] 国家创新体系建设战略研究组. 2008国家创新体系发展报告-国家创新体系研究[M]. 北京: 知识产权出版社, 2008: 174.Strategic team of national innovation system construction. Development report of national innovation system 2008 - research of national innovation system [M]. Beijing: Intellectual Property Publishing House, 2008: 174. [3] 黄洪波, 宋河发, 曲婉. 专利产业化及其评价指标体系与测度方法研究[J]. 科技进步与对策, 2011, 28(15): 110-114.Huang Hongbo, Song Hefa, Qu Wan. Patent industrialization and its evaluation index system and measure [J]. Science & Technology Progress and Policy, 2011, 28(15): 110-114. [4] 党小梅, 郑永平, 於林峰. 专利产业化的障碍性分析及对策探析[J]. 科学管理研究, 2006, 24(6): 104-107, 117.Dang Xiaomei, Zheng Yongping, Yu LinFeng. Ascertainment of drawbacks of patent industrialization and policy study [J]. Scientific Management Research, 2006, 24(6): 104-107, 117. [5] 王瑞敏, 滕青, 卢斐斐. 影响高校专利转化的因素分析和对策研究[J]. 科研管理, 2013, 34(3): 137-144.Wang Ruimin, Teng Qing, Lu Feifei. The analysis on the factors affecting the transformation of patents in university and the countermeasures [J]. Science Research Management, 2013, 34(3): 137-144. [6] 张平, 黄贤涛. 高校专利技术转化模式研究[J]. 中国高校科技, 2011(7): 13-15.Zhang Ping, Huang Xiantao. The mode of patents' technology transfer in universities [J]. Chinese University Science & Technology, 2011(7): 13-15. [7] 武建龙, 王宏起, 陶微微. 高校专利技术产业化路径选择研究[J]. 管理学报, 2012, 9(6): 884-889.Wu Jianlong, Wang Hongqi, Tao Weiwei. Path selection for industrialization of universities' patent technology [J]. Chinese Journal of Management, 2012, 9(6): 884-889. [8] 许琦, 顾新建, 陈芨熙. 基于专利引证网络的技术进化路线分析—以数据挖掘领域为例[J]. 科研管理, 2013, 34(2): 27-35.Xu Qi, Gu Xinjian, Chen Jixi. Technological evolutionary trajectories based on patent citation network - by taking data mining domain as example [J]. Science Research Management, 34(2): 27-35. [9] 马志明. 从佩奇排名到浏览排名数学为因特网建立秩序[EB/OL]. (2009-06-10)[2014-01-07], http://www.cas.cn/ys/ysjy/200906/t20090610_2330995.shtml.Ma Zhiming. Building Internet order with mathematic from page ranking to browsing ranking [EB/OL]. (2009-06-10)[2014-01-07], http://www.cas.cn/ys/ysjy/200906/t20090610_2330995.shtml. [10] Liu Yuting, Gao Bin, Liu Tieyan, et al. BrowseRank: Letting web users vote for page importance[A]//Proceedings of the 31st annual international ACM SIGIR conference on Research and development in information retrieval[C]. New York: ACM, 2008: 451-458. [11] Seneta E. Non-negative matrices and markov chains [M]. New York: Springer, 1981: 288. [12] Brin S and Page L. The anatomy of a large-scale hyper-textual web search engine[A]// Proceedings of the Seventh International World Wide Web Conference[C]. 1998: 107-117. [13] 许琦. 一种基于引证网络的文献影响因子计算方法[J]. 情报理论与实践, 2011, 34(7): 98-102.Xu Qi. A computing method for literature impact factor based on literature citation network [J]. Information Studies: Theory & Application, 2011, 34(7): 98-102. [14] 何超, 刘西林, 李佳珍. 拟Newton法在高阶矩阵中的应用——求解最大特征值及特征向量[J]. 计算机工程与应用, 2012, 48(16): 33-36, 84.He Chao, Liu Xilin, Li Jiazhen. Quasi-Newton methods for solving maximum eigenvalue and its corresponding eigenvector of high order matrix [J]. Computer Engineering and Applications, 2012, 48(16): 33-36, 84. [15] Tan Shangwang, Wang Xingke. On the largest eigenvalue of signless laplacian matrix of a graph[J]. Journal of Mathematical Research and Exposition, 2009, 29(3): 381-390. [16] United states patent and trademark office [EB/OL]. (2014-01-07) [2014-01-07]. http://patft.uspto.gov/. [17] Hall B H, Jaffe A B, Trajtenberg M. The NBER patent citation data file: Lessons, insights and methodological tools [R]. NBER Working Paper 8498, 2001. [18] Nooy W D, Mrvar A, Batagelj V. Exploratory network analysis with pajek [M]. Cambridge, MA: Cambridge University Press, 2005. [19] Assignments on the Web [EB/OL]. (2014-01-07) [2014-01-07]. http://assignments.uspto.gov/assignments/q?db=pat. [20] 邢素军. 失效专利概念辨析[J], 科技管理研究, 2013(3): 156-160.Xing Sujun. Analysis of the concept of ineffective patent [J]. Science and Technology Management Research, 2013(3): 156-160.