几种高频词阈值计算方法的实证研究——以国内高校信息素养教育研究数据为例  

An Empirical Study on Several Threshold Calculation Methods of High-Frequency Words

作　　者：王勋鸿[1] WANG Xunhong(Shandong University Library,Jinan 250100,China)

机构地区：[1]山东大学图书馆,山东济南250100

出　　处：《晋图学刊》2023年第3期30-38,共9页Shanxi Library Journal

基　　金：山东大学“新文科服务团队建设项目”(项目编号:2022CXTD02);山东大学教改项目“新文科建设背景下研究生信息素养教育改革研究”(项目编号:XYJG2020174)。

摘　　要：在图书情报学领域,关键词分析及共现分析是文献计量学中常用的一种研究方法,而高频词选取是共词分析的前提,高频词选取结果直接影响共词分析的效果,因此,高频词阈值计算方法的选择尤为重要。文章以近二十年国内高校信息素养教育研究成果为样本数据,进行基于齐普夫定律的高低频词界分法、普赖斯定律公式法、二八定律法、自定义法、词频g指数法等五种高频词阈值方法的计算,并对高频词聚类效果进行对比分析。结果发现,五种高频词阈值计算方法中,词频g指数法聚类效果最好,其他几种方法均不具有普及性和通用性,但在实际的关键词分析中,还应根据具体研究对象和研究目的,选择合适的高频词取词法。In the field of library and information science,keyword and co-occurrence analysis is an important analysis method,and the selection of high-frequency words is the premise of co-word analysis,and the result of threshold selection of high-frequency words will directly affect the effect of co-word analysis.Therefore,it is important to choose the calculation method of high-frequency word threshold.In this paper,the keywords of information literacy education of China in recent 20 years are taken as sample data,and various threshold methods of high-frequency words such as high-frequency word boundary method based on Zipf’s law,formula method based on Price’s law,80-20 Rule method,word frequency G-index method,and self-defined method are used.Among the five high-frequency word threshold calculation methods,the G index method has the best clustering effect,and other methods do not have popularity and universality.In the actual literature metrology analysis,we should choose the appropriate method according to research purpose.

关 键 词：高频词 阈值计算 聚类分析 信息素养教育 

分 类 号：G250[文化科学—图书馆学]