时间序列聚类的期刊参考文献与引证文献来源分析研究

Analysis Source Journals of References and Citation Literatures Using Time Series Data Clustering

【目的/意义】鉴于参考文献在期刊论文发表过程中的重要性以及引证文献对期刊论文的影响力,提出基于时间序列数据聚类的期刊参考文献与引证文献来源分析研究。【方法/过程】以图书情报类某重要核心期刊2010-2017年期间刊发的3500篇论文作为样本数据,利用正则表达式对非结构化的样本数据进行结构化处理,将参考文献来源期刊和引证文献来源期刊的均篇引用均值转化为时间序列数据,结合时间序列数据聚类方法分别从数值和趋势两个方面展开研究,探究发现期刊参考文献与引证文献两种来源期刊各自之间隐含的相关关系。【结果/结论】研究发现:新方法可以从时间变化的角度对来源期刊实现聚类划分,自适应地找到中心来源期刊作为簇的特征对象,其获得的结论可为目标期刊编辑部,论文读者与创作者对期刊文献的质量管理提供决策参考和理论依据。

【Purpose/significance】In view of the importance of references in the publication process of journal articles and the influence of citation literatures of the published papers,the sources Journals of references and citation literatures is studied based on the clustering of time series data.【Method/process】The 3500 papers published during the 2010-2017 periods of one key journal in the field of library and information were used as sample data.Based on regular expressions,the structured data of unstructured texts is processed by the regular expressions.The cited value per paper published on the target journal for the respective source journals of references and citation literature is changed to time series data.Combing with time series clustering method,the research is carried out in two aspects of the values and the trends.The relationships between the source journals derived from references and citation literatures are found to be implicitly related.【Result/conclusion】The study finds that the new method can cluster the time series data of the source journals in the aspect of the changeable time,and find the center journals as the feature objects of the clusters.Moreover,the obtained conclusion can provide decision-making reference and theoretical foundation to the editorial department,the readers and authors for quality management of target journal papers.