基于合理权重赋值方法选择的多因素综合评价模型——以JCR中70种人文地理期刊为例

针对地理学多因素综合评价中较少探讨指标权重赋值方法的选择,以及少有研究涉及人文地理期刊的多因素综合评价,本文设计了基于合理权重赋值方法选择的多因素综合评价模型,并将其应用于科学网数据库内的“期刊引证报告（JCR）”中分类为人文地理类的70种人文地理期刊的综合评价。该模型针对7种指标权重客观赋值方法,采用“异同比较”,分析不同方法在不同参数设置下和聚类与否情况下的权重值、评价值和评价序的变化特征与分布规律,并依据该变化特征与分布规律,实现了对各种方法的“优劣比较”和最优选择。应用该模型对JCR中70种人文地理期刊进行排序分等,其中排名第一的“全球环境变化和政策”等14种期刊为位列第一等分的高影响力期刊。本研究完善了多因素综合评价体系,也丰富了对人文地理期刊的认知。

Multi-factor comprehensive evaluation model based on the selection of objective weight assignment method

The selection of weight assignment methods in multi-factor comprehensive evaluation has been rarely discussed in geographical research. There are fewer reports concerning comprehensive evaluation of human geography journals. In this paper we propose a selection model of weight assignment methods (SWAM) for multi-factor comprehensive evaluation so as to obtain rational and appropriate weight assignment methods among a pool of candidates. Specifically, each weight assignment method SWAM conducted a 'Similar-Different' comparison to analyze the change of weight distribution, as well as the evaluation score and indicator order, according to different parameter settings and adoption of clustering. Then through the comparison of 'Good-Bad', the rational and appropriate indicator-weight assignment methods were picked up. Focusing on seven objective weight assignment methods, SWAM was finally applied for comprehensive evaluation of 70 journals of human geography sorted as geography in the database of "Journal Citation Reports" from the scientific network (Web of Science, WoB) in which the data spans from 1st January 1900 to 31st December 2012. The research results show: (1) Objective weight assignment methods in this research can be divided into five groups, namely maximizingDev, variationCoe, factorAna; greyCor-0.5; entropy; principalComAna-0.8; informationGra-0.5; (2) Change of indicator clustering parameters led to change of indicator clustering results, and has influence on weight assignment; (3) SWAM selects maximizingDev, variationCoe, factorAna-0.9 and informationGra-0.5 as the rational methods for the case study; (4) When the size of sample is relatively large, the output of grade in multi-factor comprehensive evaluation is more informative than the output of order; (5) 14 journals, led by Global Environmental Change-Human and Policy Dimensions, are marked as first-grade high-prestige group in 70 human geography journals in JCR; (6) Journal's impact factor is important, but the H-index, the distribution of authors of citing articles and the quality of citing articles, can better reflect the journal's level and international influence. This study highlights to improve the multi-factor comprehensive evaluation research and enrich the knowledge about human geography journals.