自适应聚类视角下学科领域主题分析——以物流学科为例

基于改进关键词重要性和近邻传播聚类算法,提出了一种学科领域主题分析方法,通过词频分析法和作者潜意识行为计算高频关键词权重,借助共词分析和Ochiai系数构建带权重高频关键词相似性矩阵,利用近邻传播聚类算法对学科领域主题进行自适应提取与分析。以CSSCI收录的2015—2019年物流学科领域期刊论文为研究对象,借助新方法对其进行主题分析,研究结果表明:高重要性的关键词不一定是核心主题,核心主题与同一簇内其它成员关键词最为相似。与此同时,近五年物流学科存在“降本增效”“区域物流一体化”“现代物流技术研究与创新”“物流体制改革与产业转型升级”“农村物流与农产品物流”“电子商务与跨境电商”“可持续发展理念”等七大热门主题。

Topic Analysis of Subject Fields from Perspective of Adaptive Clustering:Take Logistics Discipline As Example

In this paper,based on improved keyword importance and the affinity propagation clustering algorithm,the paper proposed a subject field topic analysis method which calculated the weight of high-frequency keywords through word frequency analysis and author subconscious behavior,built the weighted high frequency keyword similarity matrix with the help of co-word analysis and Ochiai coefficient process,and used the affinity propagation clustering algorithm to adaptively extract and analyze the topics in the subject field.Next,taking the journal papers in the logistics disciplines from 2015 to 2019 included in CSSCI as the research object,and using the methods newly proposed here to analyze the topics,we found that the high-importance keywords were not necessarily core themes,and core themes were the most similar with those in the same cluster.Another finding was that the logistics discipline in the past five years had been stage to seven major hot topics being"reducing costs and increasing efficiency","regional logistics integration","modern logistics technology research and innovation","logistics system reform and industrial transformation and upgrading","rural logistics and agricultural product logistics","Ecommerce and cross-border e-commerce",and"sustainable development concept",etc.