一种基于短文本相似度计算的知识子图融合方法  

Knowledge Subgraph Fusion Method Based on Short Text Similarity Calculation

作　　者：郑志蕴[1] 吴建萍 李钝[1] 刘允 米高扬 ZHENG Zhi-yun;WU Jian-ping;LI Dun;LIU Yun;MI Gao-yang(School of Information Engineering,Zhengzhou University,Zhengzhou 450001,China)

机构地区：[1]郑州大学信息工程学院

出　　处：《小型微型计算机系统》2020年第1期6-11,共6页Journal of Chinese Computer Systems

基　　金：国家社科基金项目(17BXW065)资助;河南省科技厅科技攻关项目(162102310616)资助;河南省教育厅教学改革研究与实践项目(32180189)资助

摘　　要：知识图谱作为语义网的数据支撑,被广泛应用于语义搜索、深度问答和在线教育等领域.知识融合是构建知识图谱的一个重要环节,将知识图中结构信息和语义信息进行融合是目前的研究热点.本文结合众包的方式,提出了一种基于短文本相似度计算的知识子图融合方法.该方法平衡各结点的结构连接和语义信息,通过学习融合权重,将高维向量转换为双邻接矩阵,得到具有高属性语义相似性的密集连接图.实验结果表明,本文提出的"群体智慧"方法能提升文本相似度计算的准确率,提高融合的质量.As the data support of semantic web,knowledge graph is widely used in semantic search,deep question answering and online education.Knowledge fusion is an important part of building knowledge graph.Fusion of structural information and semantic information in knowledge graph is a research hotspot at present.In this paper,a knowledge subgraph fusion method based on short text similarity calculation is proposed,which combines crowdsourcing.The method balances the structural connections and semantic information of each node.By learning the fusion weights,the high-dimensional vectors are transformed into double adjacency matrices,and the dense join graph with high semantic similarity of attributes is obtained.The experimental results showthat the"group wisdom"method proposed in this paper can improve the accuracy of text similarity calculation and improve the quality of fusion.

关 键 词：知识图谱 相似度计算 众包 子图融合 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]