基于带权三元闭包的知识图谱的构建方法研究  

Research on the Construction Method of Knowledge Graph Based on Weighted Triadic Closure

作　　者：孙昊天 杨良斌[1] Sun Haotian;Yang Liangbin(School of Information Technology,University of International Relations,Beijing 100091)

机构地区：[1]国际关系学院信息科技学院

出　　处：《情报杂志》2019年第6期168-173,共6页Journal of Intelligence

基　　金：中央高校基本科研业务经费“社会网络分析视角下多数据融合的知识图谱构建方法研究”(编号:3262018T30);中央在京高校人才培养共建项目:人才培养共建项目-大学生科研训练项目-信息科技学院数据科学与大数据技术科研项目

摘　　要：[目的/意义]随着时代的发展,基于关键字匹配的传统搜索方式已经不能满足人们的需求,而知识图谱正是为了满足人们在海量数据中搜索信息而产生的,目前构建社交平台的知识图谱是一个人们极为需求但研究较少的领域。[方法/过程]爬取新浪微博四个官博取其自2017年8月1日至2018年8月1日所发微博的正文及其他相关数据,并以其39 023条数据为样本。在利用共现方式标注共现次数的基础上,实现新的基于带权三元闭包来构建时政类微博的知识图谱,并改变相关参数与人工标注相较得出其最佳参数,从而生成最佳知识图谱。[结果/结论]运用以带权三元闭包为基础构建时政类微博知识图谱的方法,在一定参数约束下,可以生成符合期望的以亲密程度为关系的知识图谱。[Purpose/Significance]As the world develops, people’s needs have changed. The traditional keyword matching search method has not satisfied people’s needs. And the use of knowledge graph search has emerged, which enables people to obtain information in massive data. Currently, the knowledge graph that builds a social platform is an area that is extremely demanded but less researched.[Method/Process]The data in this article comes from Sina Weibo, including the microblog content and related data of four of the official Weibo posts from August 1, 2017 to August 1, 2018, and takes its 39 023 data as a sample. On the basis of using the co-occurrence method to mark the co-occurrence times, realizing the knowledge graph of current political Weibo based on the construction of weighted triadic closure, and the relevant parameters are compared with the manual annotation to obtain the optimal parameters. Through such a process, the best knowledge graph is finally obtained.[Results/Conclusion]With the method of constructing political microblog knowledge graph based on weighted triadic closure, and under certain parameter constraints, a knowledge graph that matches the expected degree of intimacy can be generated.

关 键 词：微博 知识图谱 三元闭包 

分 类 号：G350[文化科学—情报学]