期刊学术影响力测度指标结构关系研究——基于BP神经网络DEMATEL模型的实证

Research on the Structure of Journal Academic Influence Measurement Indexes——Based on BP Neural Network and DEMATEL Model

[目的 /意义]期刊学术影响力是学术界和期刊界关注的热点,已有许多学者对其测度指标结构关系进行了分析,但缺少从整体上对测度指标间相互作用关系的研究。[方法/过程]以632种"综合性人文、社会科学"类期刊为研究样本,构建期刊学术影响力的测度指标体系,运用BP神经网络DEMATEL模型计算各测度指标的中心度与原因度,并结合原因—结果图分析各测度指标的重要性及相互作用关系。[结果/结论]研究结果表明,该模型能较准确地反映出测度指标间的结构关系,他引影响因子和复合总被引为强驱动型指标;平均引文数和影响因子是排名前两位的驱动型指标;5年影响因子为最显著的特征型指标。

[Purpose/Significance] The journal academic influence has become the research hotspot in academic circles and periodical fields. Many scholars have analyzed the structure of journal academic influence measurement indexes,but lacked the study of the interaction between the measurement indexes on the whole. [Method/Process] In this paper,it took 632 kinds of "comprehensive humanities,social sciences" journals as the research sample. The paper construced the measurement indexes system of journal academic influence from four aspects,then used BP-DEMATEL model to calculate the degree of causal and centrality of the measurement index,and to analyze the importance and the mutual connection of each index base on the reason-result graph. [Result/Conclusion] The results showed that the non-self-citing impact factor and the composite total citation were the strong driving indexes,the top two driving indexes are the average number of citation and the impact factor,the 5-year impact factor was the most significant feature type index. This model could accurately reflect the structural relationship between indicators.