基于LDA模型的国内图书情报学研究主题发现及演化分析

The Theme Discovery and Evolution Analysis of Domestic Library and Information Science Research Based on LDA

【目的/意义】为了解近年来图书情报研究的热门主题及其演化趋势,利用LDA模型进行文本挖掘。【方法/过程】选取CNKI作为检索数据库,以2006年至2017年图书情报学领域10本核心期刊中的论文摘要作为研究数据,借助开源工具JGibbLDA构建LDA模型,运用困惑度来确定模型主题数目,根据主题-词项分布文件进行主题标识,根据文档-主题分布文件计算主题强度。【结果/结论】2006年至2017年图书情报学领域有20个研究主题,其中比较热门的主题有7个;8个主题的强度呈上升趋势,9个主题的强度呈下降趋势,3个主题的强度变化幅度较小。

【Purpose/significance】In order to understand the hot topics and evolution trends of library and information research in recent years, LDA model is used for text mining.【Method/process】Select CNKI as the search database, use the abstracts of the papers in 10 core journals in the field of library and information from 2006 to 2017 as the research data,build the LDA model with JGibbLDA, and use the confusion to determine the number of models. The term distribution file is used to identify the topic, and the topic strength is calculated based on the document-topic distribution file.【Result/conclusion】There are 20 research topics in the field of Library and Information Science from 2006 to 2017, among which there are 7 hot topics. The intensity of 8 topics is on the rise, the intensity of 9 topics is decreasing, and the intensity changing of the three topics is small.