基于LDA主题模型的科学文献主题识别:全局和学科两个视角的对比分析

Topic Identification of Scientific Literature Based on LDA Topic Model:Comparative Analysis of Two Views of Global and Discipline

[目的]利用LDA主题模型来探究全局主题和学科主题之间存在的差异。[方法]基于LDA主题模型,以国内知识流研究领域为例,在主题抽取的基础上,比较与分析全局主题和学科主题所存在的差异。[结果]全局主题和学科主题在可描述性、内容相关性、内容质量、学科分布4个方面具有显著的差异。[局限]采用经验判断选取了用于实验分析的全局热点主题和学科热点主题。

[Purpose] This paper explores the difference of global-topics and discipline-topics by using LDA（ Latent Dirichlet Allocation） topic model.[Method] Based on the LDA topic model,the paper takes the research field of knowledge flow in China as a sample to compare and analyze the difference of global-topics and discipline-topics on the basis of topic extraction.[Result]The experimental result shows the significant differences of global-topics and discipline-topics in 4 areas of descriptive,contextual,content quality,and discipline distribution.[Limitations] Hot topics from global and discipline in experiment are picked out with experience.