Incorporating message embedding into co-factor matrix factorization for retweeting prediction

Can Wang, Qiudan Li, Lei Wang, Dajun Zeng

Research output: Chapter in Book/Report/Conference proceedingConference contribution

7 Citations (Scopus)

Abstract

With the rapid growth of Web 2.0, social media has become a prevalent information sharing and spreading platform, where users can retweet interesting messages. To better understand the propagation mechanism for information diffusion, it is necessary to model the user retweeting behavior and predict future retweets. Some existing work in retweeting prediction based on matrix factorization focuses on using user-message interaction information, user information and social influence information, etc. The challenge of improving prediction performance is how to jointly perform deep representation of these information to solve the sparsity problem and then learn a more comprehensive retweeting behavior model. Inspired by word2vec and co-factor matrix factorization model, this paper proposes a hybrid model, called HCFMF, for learning users' retweeting behavior, it first computes the message content similarity by considering the message co-occurrence, the author information and word2vec based low-dimensional representation of content, then, jointly decomposes the user-message matrix and message-message similarity matrix based on a co-factorization model. We empirically evaluate the performance of the proposed model on real world weibo datasets. Experimental results show that taking the dense representation of author and content information into consideration could allow us make more accurate analysis of users' retweeting patterns. The mined patterns could serve as a feedback channel for both consumers and management departments.

Original languageEnglish (US)
Title of host publication2017 International Joint Conference on Neural Networks, IJCNN 2017 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1265-1272
Number of pages8
Volume2017-May
ISBN (Electronic)9781509061815
DOIs
StatePublished - Jun 30 2017
Event2017 International Joint Conference on Neural Networks, IJCNN 2017 - Anchorage, United States
Duration: May 14 2017May 19 2017

Other

Other2017 International Joint Conference on Neural Networks, IJCNN 2017
CountryUnited States
CityAnchorage
Period5/14/175/19/17

Fingerprint

Factorization
Feedback

Keywords

  • Co-factor matrix factorization
  • Low-dimensional representation
  • Retweeting prediction
  • Word2vec

ASJC Scopus subject areas

  • Software
  • Artificial Intelligence

Cite this

Wang, C., Li, Q., Wang, L., & Zeng, D. (2017). Incorporating message embedding into co-factor matrix factorization for retweeting prediction. In 2017 International Joint Conference on Neural Networks, IJCNN 2017 - Proceedings (Vol. 2017-May, pp. 1265-1272). [7965998] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IJCNN.2017.7965998

Incorporating message embedding into co-factor matrix factorization for retweeting prediction. / Wang, Can; Li, Qiudan; Wang, Lei; Zeng, Dajun.

2017 International Joint Conference on Neural Networks, IJCNN 2017 - Proceedings. Vol. 2017-May Institute of Electrical and Electronics Engineers Inc., 2017. p. 1265-1272 7965998.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Wang, C, Li, Q, Wang, L & Zeng, D 2017, Incorporating message embedding into co-factor matrix factorization for retweeting prediction. in 2017 International Joint Conference on Neural Networks, IJCNN 2017 - Proceedings. vol. 2017-May, 7965998, Institute of Electrical and Electronics Engineers Inc., pp. 1265-1272, 2017 International Joint Conference on Neural Networks, IJCNN 2017, Anchorage, United States, 5/14/17. https://doi.org/10.1109/IJCNN.2017.7965998
Wang C, Li Q, Wang L, Zeng D. Incorporating message embedding into co-factor matrix factorization for retweeting prediction. In 2017 International Joint Conference on Neural Networks, IJCNN 2017 - Proceedings. Vol. 2017-May. Institute of Electrical and Electronics Engineers Inc. 2017. p. 1265-1272. 7965998 https://doi.org/10.1109/IJCNN.2017.7965998
Wang, Can ; Li, Qiudan ; Wang, Lei ; Zeng, Dajun. / Incorporating message embedding into co-factor matrix factorization for retweeting prediction. 2017 International Joint Conference on Neural Networks, IJCNN 2017 - Proceedings. Vol. 2017-May Institute of Electrical and Electronics Engineers Inc., 2017. pp. 1265-1272
@inproceedings{5159927a666b4e16b8586332610a317e,
title = "Incorporating message embedding into co-factor matrix factorization for retweeting prediction",
abstract = "With the rapid growth of Web 2.0, social media has become a prevalent information sharing and spreading platform, where users can retweet interesting messages. To better understand the propagation mechanism for information diffusion, it is necessary to model the user retweeting behavior and predict future retweets. Some existing work in retweeting prediction based on matrix factorization focuses on using user-message interaction information, user information and social influence information, etc. The challenge of improving prediction performance is how to jointly perform deep representation of these information to solve the sparsity problem and then learn a more comprehensive retweeting behavior model. Inspired by word2vec and co-factor matrix factorization model, this paper proposes a hybrid model, called HCFMF, for learning users' retweeting behavior, it first computes the message content similarity by considering the message co-occurrence, the author information and word2vec based low-dimensional representation of content, then, jointly decomposes the user-message matrix and message-message similarity matrix based on a co-factorization model. We empirically evaluate the performance of the proposed model on real world weibo datasets. Experimental results show that taking the dense representation of author and content information into consideration could allow us make more accurate analysis of users' retweeting patterns. The mined patterns could serve as a feedback channel for both consumers and management departments.",
keywords = "Co-factor matrix factorization, Low-dimensional representation, Retweeting prediction, Word2vec",
author = "Can Wang and Qiudan Li and Lei Wang and Dajun Zeng",
year = "2017",
month = "6",
day = "30",
doi = "10.1109/IJCNN.2017.7965998",
language = "English (US)",
volume = "2017-May",
pages = "1265--1272",
booktitle = "2017 International Joint Conference on Neural Networks, IJCNN 2017 - Proceedings",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
address = "United States",

}

TY - GEN

T1 - Incorporating message embedding into co-factor matrix factorization for retweeting prediction

AU - Wang, Can

AU - Li, Qiudan

AU - Wang, Lei

AU - Zeng, Dajun

PY - 2017/6/30

Y1 - 2017/6/30

N2 - With the rapid growth of Web 2.0, social media has become a prevalent information sharing and spreading platform, where users can retweet interesting messages. To better understand the propagation mechanism for information diffusion, it is necessary to model the user retweeting behavior and predict future retweets. Some existing work in retweeting prediction based on matrix factorization focuses on using user-message interaction information, user information and social influence information, etc. The challenge of improving prediction performance is how to jointly perform deep representation of these information to solve the sparsity problem and then learn a more comprehensive retweeting behavior model. Inspired by word2vec and co-factor matrix factorization model, this paper proposes a hybrid model, called HCFMF, for learning users' retweeting behavior, it first computes the message content similarity by considering the message co-occurrence, the author information and word2vec based low-dimensional representation of content, then, jointly decomposes the user-message matrix and message-message similarity matrix based on a co-factorization model. We empirically evaluate the performance of the proposed model on real world weibo datasets. Experimental results show that taking the dense representation of author and content information into consideration could allow us make more accurate analysis of users' retweeting patterns. The mined patterns could serve as a feedback channel for both consumers and management departments.

AB - With the rapid growth of Web 2.0, social media has become a prevalent information sharing and spreading platform, where users can retweet interesting messages. To better understand the propagation mechanism for information diffusion, it is necessary to model the user retweeting behavior and predict future retweets. Some existing work in retweeting prediction based on matrix factorization focuses on using user-message interaction information, user information and social influence information, etc. The challenge of improving prediction performance is how to jointly perform deep representation of these information to solve the sparsity problem and then learn a more comprehensive retweeting behavior model. Inspired by word2vec and co-factor matrix factorization model, this paper proposes a hybrid model, called HCFMF, for learning users' retweeting behavior, it first computes the message content similarity by considering the message co-occurrence, the author information and word2vec based low-dimensional representation of content, then, jointly decomposes the user-message matrix and message-message similarity matrix based on a co-factorization model. We empirically evaluate the performance of the proposed model on real world weibo datasets. Experimental results show that taking the dense representation of author and content information into consideration could allow us make more accurate analysis of users' retweeting patterns. The mined patterns could serve as a feedback channel for both consumers and management departments.

KW - Co-factor matrix factorization

KW - Low-dimensional representation

KW - Retweeting prediction

KW - Word2vec

UR - http://www.scopus.com/inward/record.url?scp=85030995593&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85030995593&partnerID=8YFLogxK

U2 - 10.1109/IJCNN.2017.7965998

DO - 10.1109/IJCNN.2017.7965998

M3 - Conference contribution

AN - SCOPUS:85030995593

VL - 2017-May

SP - 1265

EP - 1272

BT - 2017 International Joint Conference on Neural Networks, IJCNN 2017 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

ER -