Loading...

Towards a Method for Inferring Trust

Akhoondi, Masoud | 2008

541 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 39010 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Habibi, Jafar
  7. Abstract:
  8. People usually use trust and reputation when they encounter uncertainty in the nature. Graphs are good tools to model and analysis of trust’s problems. If we show human and computer agents by graph’s nodes and trust values by its edges, in literature, the output graph is so-called Web of Trust. Most recently, there are three major research areas in which trust is studied. First area is policy-based trust which uses policies, digital signature and authentication methods. Second area is reputation-based trust which uses reputation and social network tools to estimate trust where past interactions or performance for an entity are combined to assess its future behavior. Third area is trust models. This area combines the two previous areas to model and define trust, its prerequisites, conditions, components, and consequences. Most of research of this thesis is related to the second area. In this thesis, an algorithm for inferring trust in social networks is presented which has two main advantages rather than the earlier ones; first it extracts trust relation from other relations and second, it makes a better balance between measures of trust estimation, applied to these networks. One of the other problems of the current approaches is to ignore trust’s context and user’s goals, which may have some contradictions. In such case, we present an algorithm to infer multi-objective trust. This method makes a suitable tradeoff among objectives. Finally, a reputation-based trust management in ad-hoc networks is presented. This approach not only deals with selfish nodes, but also copes with malicious nodes and their movements
  9. Keywords:
  10. Trust ; Reputation ; Ad Hoc Network ; Trust Web ; Multiobjective Trust

 Digital Object List

 Bookmark

No TOC