基于逆向云算法的同行评议评价模型建设探析  

Construction Analysis of Peer-Reviewed Evaluation Model Based on Inverse Cloud Algorithm

作　　者：林振亮[1] 李晓琴 燕卿[1] 郑嘉颖 黄江康[2] Lin Zhenliang;Li Xiaoqin;Yan Qing;Zheng Jiaying;Huang Jiangkang(Guangdong R&D Center for Technological Economy,Guangzhou 510070,China;Department of Science and Technology of Guangdong Province,Guangzhou 510033,China)

机构地区：[1]广东省技术经济研究发展中心,广东广州510070 [2]广东省科学技术厅,广东广州510033

出　　处：《科技管理研究》2024年第14期94-106,共13页Science and Technology Management Research

基　　金：国家自然科学基金专项项目“科学基金推动粤港澳大湾区国际科技创新中心基础研究发展的战略研究”(L2224008);广东省科技创新战略专项项目“汕头市科技专项资金绩效评价方法研究与应用分析”(STKJ202209013);广东省哲学社会科学规划学科共建项目“知识合作网络嵌入战略性产业集群创新体系发展的路径及绩效”(GD23XGL057)。

摘　　要：同行评议本质上依然是一种量化的主观定性评议,其基石是参与评议活动的每位评议者,只有做到对评议者的有效评估、监督、规范与约束,才能保证同行评议结果的客观和公正。但同行评议过程始终存在着不确定性,如何处理不确定性是同行评议需解决的难点问题。为此,在论述同行评议中评议者评价的必要性和当前发展概况的基础上,利用同行评议的评价数据分析评议者和评议影响因素,基于数据构建科学的同行评议评价模型。运用云模型理论,针对同行评议特点,选择符合正态分布统计性质、无确定度的逆向云生成算法,将方案分解为评议者评议评价与多任务协同评议评价2个层次,提出同行评议评价模型;并利用中国某省2018—2022年某类人才项目同行评议在技术、前瞻、产出、风险、经济和社会6个方向共18个评议指标的数据,通过该模型分类分层次计算出其云模型数字特征,根据数字特征构建云图和数字画像进行分析。结果表明:基于逆向云算法的同行评议评价模型对于评价评议者各方面的评议能力和整体评议效果、评价同行评议活动开展成效以及监管同行评议中的异常数据具有实操意义,能据此科学构建评议者个性化权限,能发挥评议者最大优势,并最大限度减小人为因素造成的误差,缓解当前同行评议评价面临的困境。Peer review is still essentially a quantitative subjective qualitative review,the cornerstone of which is to participate in the review activities of each reviewer,only to do an effective assessment of the reviewer,supervision,standardization,and constraints,in order to ensure that the results of the peer review of the objectivity and fairness.However,the peer review process is always uncertain,how to deal with uncertainty is a difficult problem to be solved by peer review.For this reason,on the basis of discussing the necessity of reviewer evaluation in peer review and the current development overview,the evaluation data of peer review is used to analyze the reviewers and the factors influencing the review,and a scientific peer review evaluation model is constructed based on the data.Using cloud model theory,for the characteristics of peer review,this paper selects the inverse cloud generation algorithm that conforms to the statistical nature of normal distribution and uncertainty,decomposes the program into 2 levels of reviewer review evaluation and multi-task collaborative review evaluation,and puts forward the peer review evaluation model.Further,the paper utilizes the peer review of a certain type of talent project of a certain province in China for the period of 2018-2022 in 6 directions of technology,foresight,output,risk,economy and society with a total of 18 review indicators,through the model classification and hierarchy to calculate its cloud model digital features,according to the digital features to build cloud diagrams and digital portraits for analysis.The results show that:the peer review evaluation model based on the reverse cloud algorithm has practical significance for evaluating the reviewer's ability in all aspects of the review and the overall effect of the review,evaluating the effectiveness of the peer review activities,and supervising the abnormal data in the peer review.It can scientifically build the personalized authority of the reviewer,give full play to the maximum advantage of the re

关 键 词：同行评议 评价模型 模型构建 逆向云算法 云模型 数字特征 

分 类 号：G311[文化科学] F224[经济管理—国民经济] G301