摘要
[目的 /意义]解决学术期刊多属性评价方法众多、评价结果不一致问题。[方法 /过程]提出一种基于聚类分析的多属性评价方法选取方法——聚类结果一致度筛选法。其原理是首先对原始评价指标进行聚类,然后采用可行的多属性评价方法进行评价并对评价结果进行二次聚类,最后根据评价结果聚类与原始指标聚类结果一致度的高低来选择评价方法,优先选取聚类结果一致度最高的评价方法。本文基于JCR2015数学期刊,选取11个指标,分别采用加权线性汇总、TOPSIS、VIKOR、主成分分析、调和平均进行评价,然后基于聚类结果一致度进行评价方法选取,发现调和平均的聚类一致度最高。[结果 /结论]可以采用该方法对多属性评价方法进行选择;聚类种类设置对结果影响较小;该方法具有较高的稳健性。
[ Purpose/significance ] This paper aims to solve the problems that multiple attribute evaluation method of academic journals is numerous and evaluation result is inconsistent. [ Method/process ] This paper proposes a multi- attribute evaluation method based on cluster analysis method： the cluster results consistent degree of screening method. The principle is to cluster the original evaluation index at first. Then use feasible multi-attribute evaluation method to eval- uate and the evaluation results are secondary clustering. Finally select the evaluation method according to the consistent degree of high low of evaluation results cluster and original index cluster results, and prefer to choose the evaluation meth- od that the consistent degree of cluster results is highest. This paper selects 11 indexes based on the JCR2015 journal of mathematics, and uses the weighted linear summary, TOPSIS, VIKOR, principal component analysis and harmonic evalu- ation to evaluate respectively. Then the paper selects evaluation method based on the consistent degree of cluster results, and finds that the consistent degree of harmonic average is highest. [ Result/conclusion] This method can be used to se- lect Multiple attribute evaluation methods. The cluster type of setting has little influence on the result. This method has high robustness.
作者
俞立平
Yu Liping(Zhejiang Gongshang University,School of management and E-business,Hangzhou 310018)
出处
《图书情报工作》
CSSCI
北大核心
2018年第21期80-86,共7页
Library and Information Service
基金
国家自然科学基金"基于众包的群体智慧涌现及创新效应研究"(项目编号:71472169)
教育部人文社会科学研究规划基金项目"车辆共享系统均衡优化问题机理
决策与实证研究"(项目编号:14YJA630046)研究成果之一
关键词
聚类分析
期刊评价
聚类结果一致度
评价方法选取多
属性评价
cluster analysis periodical
evaluation the consistent degree of cluster analysis the selection of evalu- ation methods
multi-attribute evaluation
