基于生成对抗网络的异质信息网络表征学习  

Heterogeneous Information Network Representation Learning Based on Generative Adversarial Network

作　　者：刘星宏 王英[1,2,3] 王鑫 兰书梅[1,2,3] LIU Xinghong;WANG Ying;WANG Xin;LAN Shumei(College of Computer Science and Technology,Jilin University,Changchun 130012,Jilin,China;College of software,Jilin University,Changchun 130012,Jilin,China;Key Laboratory of Symbol Computation and Knowledge Engineering,Ministry of Education,Jilin University,Changchun 130012,Jilin,China;College of Computer Technology and Engineering,Changchun Institute of Technology,Changchun 130012,Jilin,China)

机构地区：[1]吉林大学计算机科学与技术学院,吉林长春130012 [2]吉林大学软件学院,吉林长春130012 [3]吉林大学符号计算与知识工程教育部重点实验室,吉林长春130012 [4]长春工程学院计算机技术与工程学院,吉林长春130012

出　　处：《应用科学学报》2021年第4期532-544,共13页Journal of Applied Sciences

基　　金：国家自然科学基金(No.61872161,No.61976103);吉林省科技发展计划项目基金(No.2018101328JC,No.20200201297JC);吉林省科技厅优秀青年人才基金(No.20170520059JH);吉林省发改委项目基金(No.2019C053-8);吉林省教育厅科研项目基金(No.JJKH20191257KJ)资助。

摘　　要：鉴于传统的异质信息网络通常存在的高维稀疏性缺点,首先提出将异质信息网络的高维顶点嵌入低维向量空间的无监督学习模型——基于生成对抗网络的异质网络表征学习(heterogeneous network representation learning based on generative adversarial network,HNRL-GAN)模型;然后分析HNRL-GAN模型中的不足之处,进一步提出改进后的基于生成对抗网络的增强版异质网络表征学习(heterogeneous network representation learning based on generative adversarial network plus plus,HNRL-GAN++)模型;最后分别在DBLP、Yelp、Aminer等数据集中使用HNRL-GAN模型和HNRL-GAN++模型进行节点分类和节点聚类等实验以测试模型的有效性。实验结果表明:1)HNRL-GAN模型和HNRL-GAN++模型都实现了将异质信息网络中的高维稀疏节点表示为低维稠密向量这一目标;2)相较于HNRL-GAN模型,HNRL-GAN++模型在保留高维空间中网络结构信息和语义信息等方面拥有更好的性能。In view of the high-dimensional sparsity shortcomings of traditional heterogeneous information networks,we firstly proposed an unsupervised learning model--heterogeneous network representation learning based on generative adversarial network(HNRLGAN)that embeds the high-dimensional vertices of heterogeneous information networks into low-dimensional vector spaces.Secondly,having analyzed the shortcomings of HNRLGAN,we proposed an improved model,called as heterogeneous network representation learning based on generative adversarial network plus plus(HNRL-GAN++).Finally,we used HNRL-GAN and HNRL-GAN++in three data sets,including DBLP,Yelp,and Aminer,to perform node classification and node clustering for testing the effectiveness of the two models.Experimental results show that:1)Both HNRL-GAN and HNRLGAN++achieve the goal of representing high-dimensional sparse nodes in heterogeneous information networks as low-dimensional dense vectors;2)Compared with HNRL-GAN,HNRL-GAN++has better performance in retaining network structure information and semantic information in high-dimensional space.

关 键 词：异质信息网络 生成对抗网络 网络表征学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]