Reputation Description and Interpretation

Abstract

Reputation is an opinion held by others about a particular person, group, organisation, or resource. As a tool, reputation can be used to forecast the reliability of others based on their previous actions, moreover, in some domains it can even be used to estimate trustworthiness. Due to the large scale of virtual communities it is impossible to maintain a meaningful relationship with every member. Reputation systems are designed explicitly to manufacture trust within a virtual community by recording and sharing information regarding past interactions. Reputation systems are becoming increasingly popular and widespread, with the information generated varying considerably between domains. Currently, no formal method to exchange reputation information exists. However, the OpenRep framework, currently under development, is designed to federate reputation information, enabling the transparent exchange of information between reputation systems. This thesis presents a reputation description and interpretation system, designed as a foundation for the OpenRep framework. The description and interpretation system focuses on enabling the consistent and reliable expression and interpretation of reputation information across heterogeneous reputation systems. The description and interpretation system includes a strongly typed language, a verification system to validate usage of the language, and a XML based exchange protocol. In addition to these contributions, three case studies are presented as a means of generating requirements for the description and interpretation system, and evaluating the use of the proposed system in a federated reputation environment. The case studies include an electronic auction, virtual community and social network based relationship management service.