首页> 外文会议>Mexican international conference on artificial intelligence >Hermitian Laplacian Operator for Vector Representation of Directed Graphs: An Application to Word Association Norms
【24h】

Hermitian Laplacian Operator for Vector Representation of Directed Graphs: An Application to Word Association Norms

机译:有向图的矢量表示的埃尔米特Laplacian运算符:在单词联想规范中的应用

获取原文

摘要

In this paper, we propose a spectral method for the analysis of directed graphs. For this purpose, a Hermitian Laplacian operator is proposed, that defines interesting properties for the embedding of a graph into a vector space. We use the notions of the Hermitian Laplacian operator to embed a directed graph structure, build over corpura of Word Association Norms, into a vector space. We show that the Hermitian Laplacian operator has advantages over a traditional Laplacian operator when the original structure of the graph is directed. Moreover, we compare the lexical relations obtained by a WAN graph with the connections the Hermitian Laplacian operator establishes between the words of the corpora.
机译:在本文中,我们提出了一种用于有向图分析的频谱方法。为此,提出了一种Hermitian Laplacian运算符,该运算符定义了将图嵌入向量空间的有趣属性。我们使用Hermitian Laplacian运算符的概念将有向图结构嵌入到单词关联规范的语料库之上,并将其嵌入到向量空间中。我们表明,当有向图的原始结构被定向时,埃尔米特Laplacian算子比传统Laplacian算子具有优势。此外,我们将WAN图获得的词汇关系与Hermitian Laplacian运算符在语料库词之间建立的连接进行了比较。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号