Trust Evaluation and Spam Detection on Online Social Networks

  • Yang, Jian (Primary Chief Investigator)
  • Sheng, Michael (Primary Chief Investigator)
  • Wu, Jia (Chief Investigator)
  • Collins, Alex (Partner Investigator)
  • Nepal, Surya (Partner Investigator)
  • Paris, Cécile (Partner Investigator)

Project: Other

Project Details


As online social networks (OSNs) become a popular media for sharing thoughts, opinions, and experiences. We see news travels fast on Tweeter, some are true, some are deceptive, and some are absolutely false. Therefore the trust of information and of users is becoming increasingly a concern and a challenge to the society. OSN users may have a broad range of interactions with other users, applications, and systems and leave behind informative and rich traces that, if captured and interpreted correctly, can be used as the basis to evaluate the trustworthiness of users and/or information pieces on OSNs. Trust evaluation is normally context dependent [1]. This can include rich provenance data such as: the creator of the content; the time, place, and the reason for the creation; and how it is related with other information. Due to this rich context and the complex interactions among people around the information created, the task of trust evaluation and spam detection is challenging in terms of accuracy, relevancy, and consistency.
In order to evaluate the trustworthy of a piece of information and people available on OSNs, our initial investigation indicates that we need to study both the information network and the user network and build relationships between these two networks. Information network refers to linkage between relevant information items, while user network refers to the linkage between relevant users who create, consume, propagate, or modify the information. Importantly, user networks are dynamic, and this aspect needs to be taken into account.
Short titleTrust
Effective start/end date3/12/1831/08/21