Continuous ratings in discrete Bayesian reputation systems

Abstract

Reputation systems take as input ratings from members in a community, and can produce measures of reputation, trustworthiness or reliability of entities in the same community. Binomial and multinomial Bayesian reputation systems are discrete in nature meaning that they normally take discrete ratings such as "average" or "good" as input. However, in many situations it is natural to provide input ratings to reputation systems based on continuous measures. This paper describes the principles of discrete Bayesian reputation systems, and how continuous measures can provide input ratings to such systems. The method is based on fuzzy set membership functions. © 2008 International Federation for Information Processing.