基于LDA的信息资源挖掘与可视化研究  

Research on Information Resources Mining and Visualization Based on LDA

作　　者：丁玲[1] 叶佳鑫 曾婷[1] DING Ling;YE JiaXin;ZENG Ting(Wuhan Land Resource and Urban Planning Information Centre, Wuhan 430014, China;School of Information Management, Central China Normal University, Wuhan 430079, China)

机构地区：[1]武汉市国土资源和规划信息中心,武汉430014 [2]华中师范大学信息管理学院,武汉430079

出　　处：《数字图书馆论坛》2019年第2期29-36,共8页Digital Library Forum

摘　　要：LDA(Latent Dirichet Allocation)是一种从文档资源中抽取主题的概率模型,将其用于文档的主题提取通常具有不错的效果。档案信息资源是一种具有较高利用价值的文档资源,但其目前存在碎片化、建设不足等问题。基于此,本文将LDA与聚类、层次空间构建技术相结合应用于档案信息资源建设,并进行实证研究。从实验结果来看,将LDA应用于档案信息资源建设可以挖掘资源间的隐含联系,明确资源间的等级层次,并有助于信息资源的可视化展示。LDA is a probabilistic model for extracting topics from document resources, and it is usually effective for extracting topics from documents. Archival information resources are of high utilization value, but there are problems such as fragmentation and insufficient construction. Based on this, this paper applies LDA, clustering and hierarchical space construction technology to the construction of archival information resources, and conducts an empirical study. From the experimental results, the application of LDA to the construction of archival information resources can excavate the hidden connections between resources, clarify the hierarchy of resources, and contribute to the visual display of information resources.

关 键 词：信息资源建设 主题提取 聚类 层次空间 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]