融合多维特征的高校专利价值分级方法及其实证研究

An Empirical Study on Value-based Grading Method for University Patents with Multidimensional Features

客观评估高校专利价值,准确预测高校专利的技术转移潜力,有助于推动高校技术成果与市场对接,促进科研成果转化成生产力,对企业精准引进高价值专利、促进技术升级具有意义。文章从高校专利价值的3个重要来源出发,融合专利文本语义特征、专利自身属性特征、发明人特征,采用机器学习算法构建高校专利价值评估模型,预测专利发生技术转移的概率,并划定阈值将专利价值的粒度细化为10个等级,以云计算领域为例展开实证。结果表明:文章构建的模型与现有研究相比,AUC值提升22个百分点,显著提升高校专利价值评估结果的准确性和科学性。对专利价值的分级结果表明,A级别专利仅占3.24%,大多数高校专利的技术转移概率处于0.3-0.5,整体呈现明显的右偏态分布。高校专利价值分级方法能够有效预测技术转移潜力,可为高校合理进行专利运营、企业精准定位高价值专利提供决策参考,也为高校专利价值评估提供了新思路。

Objectively evaluating the value of university patents and accurately predicting the technology transfer potential of university patents will not only help promote closer linkage between university technological achievements and the market,and expedite the transformation of research results,but also have positive significance for enterprises to accurately introduce high-value patents and promote technology upgrading.Based on the three important origins of university patent value,the semantic characteristics of patent text,the attribute characteristics of patent itself and the characteristics of inventor are incorporated.This paper develops an evaluation model for university patent,using machine learning algorithm,to predict the probability of technology transfer from patents,and subdivide the granularity of patent value into ten grades by setting certain thresholds.An empirical study is carried out in the field of cloud computing.The findings suggest that the AUC value of the model increases by 22 percentage compared with those in other researches,which has made the patent value evaluation in colleges and universities more accurately and scientifically.The grading results of patent value show that grade A patents account for only 3.24%,and the technology transfer rate of most university patents is between 0.3 and 0.5,with an obvious right-skewed distribution as a whole.The value-based grading method can effectively predict the potentiality of technology transfer,provide reference for universities to maintain reasonable patent operation and for enterprises to accurately find out the high-value patents,as well as giving fresh ideas on the assessment of the value of university patents.