Universal network representation for heterogeneous information networks

Ruiqi Hu, Celina Ping Yu, Sai Fu Fung, Shirui Pan, Haishuai Wang, Guodong Long

Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

Abstract

Network representation aims to represent the nodes in a network as continuous and compact vectors, and has attracted much attention in recent years due to its ability to capture complex structure relationships inside networks. However, existing network representation methods are commonly designed for homogeneous information networks where all the nodes (entities) of a network are of the same type, e.g., papers in a citation network. In this paper, we propose a universal network representation approach (UNRA), that represents different types of nodes in heterogeneous information networks in a continuous and common vector space. The UNRA is built on our latest mutually updated neural language module, which simultaneously captures inter-relationship among homogeneous nodes and node-content correlation. Relationships between different types of nodes are also assembled and learned in a unified framework. Experiments validate that the UNRA achieves outstanding performance, compared to six other state-of-the-art algorithms, in node representation, node classification, and network visualization. In node classification, the UNRA achieves a 3% to 132% performance improvement in terms of accuracy.

Original languageEnglish
Title of host publicationInternational Joint Conference on Neural Networks (IJCNN) 2017
Subtitle of host publicationAnchorage, Alaska, USA May 14 – May 19, 2017
EditorsChrisina Jayne
Place of PublicationPiscataway NJ USA
PublisherIEEE, Institute of Electrical and Electronics Engineers
Pages388-395
Number of pages8
ISBN (Electronic)9781509061815, 9781509061822
ISBN (Print)9781509061839
DOIs
Publication statusPublished - 2017
Externally publishedYes
EventIEEE International Joint Conference on Neural Networks 2017 - Anchorage, United States of America
Duration: 14 May 201719 May 2017
https://web.archive.org/web/20170502003739/http://www.ijcnn.org/

Conference

ConferenceIEEE International Joint Conference on Neural Networks 2017
Abbreviated titleIJCNN 2017
CountryUnited States of America
CityAnchorage
Period14/05/1719/05/17
Internet address

Cite this

Hu, R., Yu, C. P., Fung, S. F., Pan, S., Wang, H., & Long, G. (2017). Universal network representation for heterogeneous information networks. In C. Jayne (Ed.), International Joint Conference on Neural Networks (IJCNN) 2017: Anchorage, Alaska, USA May 14 – May 19, 2017 (pp. 388-395). [7965880] Piscataway NJ USA: IEEE, Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/IJCNN.2017.7965880
Hu, Ruiqi ; Yu, Celina Ping ; Fung, Sai Fu ; Pan, Shirui ; Wang, Haishuai ; Long, Guodong. / Universal network representation for heterogeneous information networks. International Joint Conference on Neural Networks (IJCNN) 2017: Anchorage, Alaska, USA May 14 – May 19, 2017. editor / Chrisina Jayne. Piscataway NJ USA : IEEE, Institute of Electrical and Electronics Engineers, 2017. pp. 388-395
@inproceedings{51dddcc854604b49bb5525efc77e37d7,
title = "Universal network representation for heterogeneous information networks",
abstract = "Network representation aims to represent the nodes in a network as continuous and compact vectors, and has attracted much attention in recent years due to its ability to capture complex structure relationships inside networks. However, existing network representation methods are commonly designed for homogeneous information networks where all the nodes (entities) of a network are of the same type, e.g., papers in a citation network. In this paper, we propose a universal network representation approach (UNRA), that represents different types of nodes in heterogeneous information networks in a continuous and common vector space. The UNRA is built on our latest mutually updated neural language module, which simultaneously captures inter-relationship among homogeneous nodes and node-content correlation. Relationships between different types of nodes are also assembled and learned in a unified framework. Experiments validate that the UNRA achieves outstanding performance, compared to six other state-of-the-art algorithms, in node representation, node classification, and network visualization. In node classification, the UNRA achieves a 3{\%} to 132{\%} performance improvement in terms of accuracy.",
author = "Ruiqi Hu and Yu, {Celina Ping} and Fung, {Sai Fu} and Shirui Pan and Haishuai Wang and Guodong Long",
year = "2017",
doi = "10.1109/IJCNN.2017.7965880",
language = "English",
isbn = "9781509061839",
pages = "388--395",
editor = "Jayne, {Chrisina }",
booktitle = "International Joint Conference on Neural Networks (IJCNN) 2017",
publisher = "IEEE, Institute of Electrical and Electronics Engineers",
address = "United States of America",

}

Hu, R, Yu, CP, Fung, SF, Pan, S, Wang, H & Long, G 2017, Universal network representation for heterogeneous information networks. in C Jayne (ed.), International Joint Conference on Neural Networks (IJCNN) 2017: Anchorage, Alaska, USA May 14 – May 19, 2017., 7965880, IEEE, Institute of Electrical and Electronics Engineers, Piscataway NJ USA, pp. 388-395, IEEE International Joint Conference on Neural Networks 2017, Anchorage, United States of America, 14/05/17. https://doi.org/10.1109/IJCNN.2017.7965880

Universal network representation for heterogeneous information networks. / Hu, Ruiqi; Yu, Celina Ping; Fung, Sai Fu; Pan, Shirui; Wang, Haishuai; Long, Guodong.

International Joint Conference on Neural Networks (IJCNN) 2017: Anchorage, Alaska, USA May 14 – May 19, 2017. ed. / Chrisina Jayne. Piscataway NJ USA : IEEE, Institute of Electrical and Electronics Engineers, 2017. p. 388-395 7965880.

Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

TY - GEN

T1 - Universal network representation for heterogeneous information networks

AU - Hu, Ruiqi

AU - Yu, Celina Ping

AU - Fung, Sai Fu

AU - Pan, Shirui

AU - Wang, Haishuai

AU - Long, Guodong

PY - 2017

Y1 - 2017

N2 - Network representation aims to represent the nodes in a network as continuous and compact vectors, and has attracted much attention in recent years due to its ability to capture complex structure relationships inside networks. However, existing network representation methods are commonly designed for homogeneous information networks where all the nodes (entities) of a network are of the same type, e.g., papers in a citation network. In this paper, we propose a universal network representation approach (UNRA), that represents different types of nodes in heterogeneous information networks in a continuous and common vector space. The UNRA is built on our latest mutually updated neural language module, which simultaneously captures inter-relationship among homogeneous nodes and node-content correlation. Relationships between different types of nodes are also assembled and learned in a unified framework. Experiments validate that the UNRA achieves outstanding performance, compared to six other state-of-the-art algorithms, in node representation, node classification, and network visualization. In node classification, the UNRA achieves a 3% to 132% performance improvement in terms of accuracy.

AB - Network representation aims to represent the nodes in a network as continuous and compact vectors, and has attracted much attention in recent years due to its ability to capture complex structure relationships inside networks. However, existing network representation methods are commonly designed for homogeneous information networks where all the nodes (entities) of a network are of the same type, e.g., papers in a citation network. In this paper, we propose a universal network representation approach (UNRA), that represents different types of nodes in heterogeneous information networks in a continuous and common vector space. The UNRA is built on our latest mutually updated neural language module, which simultaneously captures inter-relationship among homogeneous nodes and node-content correlation. Relationships between different types of nodes are also assembled and learned in a unified framework. Experiments validate that the UNRA achieves outstanding performance, compared to six other state-of-the-art algorithms, in node representation, node classification, and network visualization. In node classification, the UNRA achieves a 3% to 132% performance improvement in terms of accuracy.

UR - http://www.scopus.com/inward/record.url?scp=85031033901&partnerID=8YFLogxK

U2 - 10.1109/IJCNN.2017.7965880

DO - 10.1109/IJCNN.2017.7965880

M3 - Conference Paper

SN - 9781509061839

SP - 388

EP - 395

BT - International Joint Conference on Neural Networks (IJCNN) 2017

A2 - Jayne, Chrisina

PB - IEEE, Institute of Electrical and Electronics Engineers

CY - Piscataway NJ USA

ER -

Hu R, Yu CP, Fung SF, Pan S, Wang H, Long G. Universal network representation for heterogeneous information networks. In Jayne C, editor, International Joint Conference on Neural Networks (IJCNN) 2017: Anchorage, Alaska, USA May 14 – May 19, 2017. Piscataway NJ USA: IEEE, Institute of Electrical and Electronics Engineers. 2017. p. 388-395. 7965880 https://doi.org/10.1109/IJCNN.2017.7965880