摘要
[目的 /意义]对以图书为代表的多主题长文档进行文档内层次主题分析及组织,为用户提供细粒度的挖掘结果,以帮助用户了解图书主题,并快速理解图书内部主题的结构与联系。[方法 /过程]在利用层次主题模型hLDA及上下文信息构建图书内部主题层次组织模型并进行模型实现的基础上,设计实验对模型进行评估。[结果 /结论]实验结果表明,基于hLDA的图书内部主题层次组织具有更高的查全率和查准率。
[ Purpose/significance ] This paper analyzes and organizes hierarchical topic texts in multi -topic long documents which represented by hooks, and offers fine - granularity mining results for users to help them understand the topic of a book and quickly understand the structure and relationship of topics within the book. [ Method/process ] Firstly, hierarchical topic model (hLDA) and context information are applied to build hierarchical topic organization model within the book and its prototype system is implemented. Secondly, an experiment is designed to evaluate this model. [ Result/conclusion] The experiment results prove that the internal hierarchical topic organization model of a book will promote the recall and the precision.
出处
《图书情报工作》
CSSCI
北大核心
2016年第18期140-148,共9页
Library and Information Service
基金
国家自然科学基金项目"图书层次主题自动标引研究"(项目编号:71303089)
华中师范大学2016年校级教学研究项目"信息管理类‘知识主题-课程’体系网络构建研究"(项目编号:201623)研究成果之一