A Method for Assessing the Trustworthiness of an Entity by Cooperating Authorities

Mathias Kohler

16 October 2005

Abstract: In this thesis a Model of Trust based on Bayesian Networks is introduced. The model determines the trustworthiness of a new or hitherto unknown entity by utilizing information exchanged with cooperative authorities. Results from different trust models using individual evaluation techniques to assess the trustworthiness of an entity can only be utilized using specific conversion methods or not at all. However, utilizing results from other authorities is always needed if an authority cannot evaluate an entity on its own. This would be the case, for instance, if no previous interactions with an entity have taken place and hence no experience is available for an evaluation of an entity's trustworthiness. This trust model addresses this problem and introduces a method to enable information exchange with cooperative authorities to assess a new or unknown entity, even if they use different evaluation techniques. The method is based on the exchange of trust relations. This means, rather than using absolute values reflecting the degree of confidence in an entity, this model makes use of a relative notion of trust. An entity is hereby seen as being equally or more trustworthy than another entity. The trust relations are modelled using the structure and inference algorithms of Bayesian Networks. By finally evaluating them, a new or unknown entity's trustworthiness is assessed.

Keyword(s): trust model ; Bayesian Network ; Access Control Models and Languages

Published in: Mathias Kohler, "A Method for Assessing the Trustworthiness of an Entity by Cooperating Authorities," Master thesis in Communication and Media Engineering, Offenburg, Germany, University of Applied Sciences, Sep, 2005, pp.66. :

The record appears in these collections:
Access Control Models and Languages

 Record created 2009-04-27, last modified 2013-05-22

Transfer from CDS 0.99.7:
Download fulltext

Rate this document:

Rate this document:
(Not yet reviewed)