基于改进谱聚类算法的生物制药产业专利价值评估研究  

Research on Patent Value Evaluation of Biopharmaceutical Industry Based on Improved Spectral Clustering Algorithm

作　　者：朱文韵 郭晴晴 Zhu Wenyun;Guo Qingqing(School of Economics and Management,Shanghai Institute of Technology,Shanghai 200235,China)

机构地区：[1]上海应用技术大学经济与管理学院,上海200235

出　　处：《科技管理研究》2023年第20期57-64,共8页Science and Technology Management Research

摘　　要：有效评估药物专利价值有必要考虑制药基础技术细节以及新药专利保护期限较长的特殊性等有关实际,同时,利用机器学习方法开展专利价值评估的研究仍有待进一步完善,因此,针对生物制药产业专利价值评估准确性问题,结合产业技术因素及其专利特点,以及专利价值评估的共性指标和生物制药产业特征与专利技术特点的个性指标,引入自编码器(AE)模型和谱聚类算法(SC)构建专利价值评估算法模型,以药智专利通数据库的相关专利数据为样本进行实证分析,通过提取专利指标特征、专利聚类来进行专利价值评估,并运用支持向量机方法对专利价值进行分类,以验证AE-SC评估模型的有效性。结果表明:AE-SC评估模型通过自编码器提取专利特征后的专利价值聚类准确度优于谱聚类和传统K-means聚类;专利存在年数、药物专利类型、适应证类别等是评价生物制药产业专利价值必要考虑因素。In order to effectively evaluate the value of pharmaceutical patents,it is necessary to consider the basic technical details of pharmaceuticals and the peculiarity of the long patent protection period of new drugs.At the same time,the research on patent value assessment by using machine learning method still needs to be further improved.Therefore,in view of the accuracy of patent value assessment in the bio-pharmaceutical industry,combining industrial technology factors and its patent characteristics,as well as the common index of patent value assessment and the individual index of bio-pharmaceutical industry characteristics and patent technology characteristics,the auto encoder(AE)model and spectral clustering algorithm(SC)are introduced to build the patent value assessment algorithm model,and the relevant patent data of the Yaozhi patent database are used as samples for empirical analysis.The patent value assessment is carried out by extracting the characteristics of patent indicators and conducting patent clustering,support vector machine method is used to classify the patent value to verify the validity of the evaluation model.The results show that the accuracy of AE-SC evaluation model after extracting patent features by auto encoder is better than spectral clustering and traditional K-means clustering;the number of years of patent existence,the type of drug patent and the type of indication are the necessary factors to evaluate the patent value of bio-pharmaceutical industry.

关 键 词：药物专利价值评估 技术因素 专利特点 自编码器 谱聚类 支持向量机 生物制药产业 

分 类 号：G306[文化科学] F224[经济管理—国民经济] G301