基于智能主题图的科技文献细粒度知识组织模型  

A Fine-Grained Knowledge Organization Model for Scientific Literature Based on Intelligent Topic Graphs

作　　者：秦春秀[1] 郑梦悦 马续补[1] 赵捧未[1] Qin Chunxiu;Zheng Mengyue;Ma Xubu;Zhao Pengwei(School of Economics and Management,Xidian University,Xi'an 710071)

机构地区：[1]西安电子科技大学经济管理学院,西安710071

出　　处：《情报杂志》2021年第11期169-175,135,共8页Journal of Intelligence

基　　金：国家自然科学基金项目“知识社区中的资源语义空间及其检索研究”(编号:71573199);陕西省软科学一般项目“陕西省科技信息服务的新精矿发展模式及策略研究”(编号:2021RPM136);陕西省教育厅重点科学研究计划“‘丝绸之路经济带’区域信息化协调发展及其影响因素研究”(编号:20JZ056)。

摘　　要：[研究目的]大多数科技文献检索系统仍采用关键词方法对篇章级的科技文献进行描述和组织,这种知识组织方式缺乏对文献内容语义的丰富描述与组织,不能满足用户精准化的知识需求。为了实现对每篇科技文献内部内容的语义导航,向用户提供精准化的检索内容,对科技文献内部知识进行细粒度描述与组织变得更加迫切。[研究方法]结合智能主题图结构,构建了一个基于智能主题图的科技文献细粒度知识组织模型。该模型在分析文献集合的内部内容特征基础上,抽取文献内部的主题以及知识单元,将文献集合划分为资源层、知识单元层、主题层以及聚类层,并采用全信息相似度方法计算主题及主题之间的关联、知识单元及知识单元间的语义关联度。[研究结论]以《中国图书馆学报》中的部分文献为数据集进行实验,以图形化的方式对该模型进行了展示,并与传统检索结果进行比较,结果表明,本研究提出的科技文献细粒度知识组织模型能对知识单元进行较为准确导航与检索。[Research purpose]Most scientific and technological literature retrieval systems still use the method of describing and organizing literature by the unit of paper,which lacks the rich description and organization of the semantic content of literature,and cannot meet the precise knowledge needs of users.In order to realize the knowledge navigation of the internal content of each scientific and technological literature and provide users with accurate retrieval content,it becomes more urgent to organize the internal knowledge of scientific and technological literature with fine-grained knowledge.[Research method]Based on the analysis of the internal knowledge of sci-tech literature and the structure of intelligent topic map,this paper constructs a fine-grained knowledge organization model of sci-tech literature based on intelligent topic map.Based on the analysis of the internal content of literature collection,the model extracts theme of the literature and the internal knowledge units,and divides the literature collection into resources layer,knowledge unit layer,subject layer and clustering layer.Full information similarity method is used to compute the link between the themes,and the semantic correlation degree between the knowledge units.[Research conclusion]Some literatures in Journal of Library Science in China are used as data sets to demonstrate the model in a graphical way,and compared with the traditional retrieval results.The results show that the fine-grain knowledge organization model of scientific and technological literature proposed in this paper can accurately navigate and retrieve knowledge units.

关 键 词：科技文献 智能主题图 知识单元 知识组织 

分 类 号：G350[文化科学—情报学]