考虑共识度的学者学术影响力组合评价研究

Research on Combination Evaluation of Academic Influence of Scholars Considering Consensus Degree

[目的/意义]针对多种学者学术影响力综合评价方法结论非一致性的问题,提出一种新的考虑共识度的学者学术影响力组合评价模型。[方法/过程]首先,定义本年度总被引次数、上年度总被引次数和前年度总被引次数的概念,结合H指数和总被引次数构建一套学者学术影响力评价指标体系;其次,利用熵权TOPSIS方法、VIKOR方法、Heronian算子、DEMATEL方法和幂算子对学者学术影响力进行综合评价,计算5种不同的学者学术影响力综合评价结果;最后,设计考虑共识度的信息融合方法,计算学者学术影响力组合评价结果。[结果/结论]选取26位国际模糊决策领域专家为研究对象进行实证分析,分析表明,该模型不仅可以发挥各种综合评价方法的优点和弥补各种综合评价方法的缺点,可以避免人为选取综合评价方法进行操纵的可能性,还可以使得最终的组合评价结果相对更为可靠,因此能够得到较为客观公正的学者学术影响力排名。

[Purpose/significance] Aiming at the problem that the results of several comprehensive evaluation methods of academic influence of scholars are inconsistent, this paper proposes a novel combination evaluation model of academic influence of scholars considering consensus degree.[Method/process] Firstly, the concepts of this year’s total citations, last year’s total citations, and previous year’s total citations are defined, on the basis of which the index system of academic influence of scholars that contain H index, total citations, and these concepts is constructed.Then, the entropy weight-based Technique for Order Preference by Similarity to an Ideal Solution(EW-TOPSIS) method, Vlse Kriterijuska Optimizacija I Komoromisno Resenje(VIKOR) method, Heronian operator, Decision-making Trial and Evaluation Laboratory(DEMATEL) method, and power operator are utilized to calculate 5 different comprehensive evaluation results of academic influence of scholars.Finally, the information fusion method considering consensus degree is designed to determine the combination evaluation results of academic influence of scholars.[Result/conclusion] Taking 26 experts of international fuzzy decision-making field as the object, the empirical analysis shows that this model can get an objective and fair rank of academic influence of scholars owing to not only can this model consider the advantages and disadvantages of each comprehensive evaluation method, and avoid the adverse influence of the random selection of comprehensive evaluation methods, but this model can also get more reliable combination evaluation results.