科技评价指标权重分类及对评价的影响研究

Research on Weighting Characteristics of Scientific and Technical Evaluation Index Data

[目的/意义]本文认为在科技评价指标中隐含3种权重:实际权重、自然权重、数据分布权重,为了分析它们之间的关系及对评价方法选择产生的影响。[方法/过程]本文以JCR2016数学期刊为例,同时选取主成分分析和TOPSIS作为非线性评价方法的代表进行对比分析,并提出采用动态最大均值标准化方法消除自然权重。[结果/结论]评价指标自然权重影响线性评价结果;评价指标数据分布权重主要影响非线性评价结果;不同非线性评价方法也会影响非线性评价的实际权重;指标权重特性影响评价方法的选择,线性评价可以消除自然权重,而且不存在数据分布权重问题,非线性评价数据分布权重无法消除,因而从评价方法角度,线性评价更为合理。设计权重与实际权重力求一致是选择评价方法的重要依据。

[Purpose/Significance]There are three kinds of weights in the evaluation index data of science and technology.One is reality weight,the second is natural weight,the third is data distribution weight.To analyze their relationship is very important.[Method/Process]The paper taking the JCR2016 Mathematical Journal as example,compared and analyzed principal component analysis and TOPSIS as the representatives of nonlinear evaluation methods and propose standardization method of approximate dynamic maximum mean to eliminate the natural weights.[Result/Conclusion]It showed that the natural weights of evaluation indexes affected the results of linear evaluation;The weights of evaluation index data mainly affected the nonlinear evaluation results;Different nonlinear evaluation methods also affected the actual weights of nonlinear evaluation;the weight characteristics of index data affected the choice of evaluation methods;The linear evaluation can eliminate the natural weights,and there is no problem of the weight of the data distribution;The weight of nonlinear evaluation data distribution can not be eliminated,so the linear evaluation is more reasonable in the view of evaluation method.