Privacy Protection in Trust Models for Agent Societies

Abstract

Abstract. In this paper we have motivated the use of privacy-protection measures in trust models, both in conscious exchanges of opinions and in an unconscious way when security attacks take place. Most of the privacy dimensions are concerned into trust communications. In particular we define the privacy rights that these trusting communications must legally be guaranteed. From them, we describe additional message exchanges that, acting as control mechanisms, would be required to exercise such rights. Furthermore, we also enumerated the corresponding privacy violations that would have taken place if these control mechanisms were ignored. From the possible existence of privacy violations, regulatory structures may establish what agents are allowed and forbidden to do according to the legal privacy rights. We have applied the control mechanisms as additional message exchanges to a particular application domain (the Agent Trust and Reputation testbed) implemented as JADE interaction protocols, and finally we plan to define an Electronic Institution that would rule the corresponding norms and violations to such control using the Islander specification tool.