Health-related spammer detection on chinese social media

Xinhuan Chen, Yong Zhang, Jennifer Xu, Chunxiao Xing, Hsinchun Chen

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


Weibo (Chinese microblog) has become a popular social media platform for users to share health-related information. However, illegitimate users or spammers often generate and spread false or misleading health information so as to advertise and attract more attention. To address this issue, we propose a healthrelated spammer detection approach on Chinese social media. Our approach is a deep belief network (DBN) based model incorporating a comprehensive feature set, including burstiness-based features, profile-based features, and content-based features, to identify spammers who spread misleading health-related information. Especially, we create a medical and health domain lexicon to better extract content-based features. The experimental results show the approach achieves an F1 score of 86% in detecting spammer and significantly outperforms the benchmark methods using baseline features.

Original languageEnglish (US)
Title of host publicationSmart Health - International Conference, ICSH 2015, Revised Selected Papers
EditorsHsinchun Chen, Daniel Dajun Zeng, Xiaolong Zheng, Scott J. Leischow
Number of pages12
ISBN (Print)9783319291741
StatePublished - 2016
EventInternational Conference for Smart Health, ICSH 2015 - Phoenix, United States
Duration: Nov 17 2015Nov 18 2015

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


OtherInternational Conference for Smart Health, ICSH 2015
Country/TerritoryUnited States


  • Chinese
  • Deep belief network
  • Health
  • Spammer detection
  • Weibo

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science


Dive into the research topics of 'Health-related spammer detection on chinese social media'. Together they form a unique fingerprint.

Cite this