Accumulative systems

Calculate the reputation of an agent as the sum of all given ratings. e. g. e-bay

Average systems

The reputation of an agent is computed as the average of all ratings that the agent has received. e.g. Jurca and Faltings system; Yu and Singh system

Blurred Systems

Provide a reputation metric with an unspecified time-dependent weight function, where old ratings lose their influences on the current reputation over time. An agent behaves like it did the last time. e.g. Only last system

Beta System

It is designed with the purpose of predicting statistically an agent’s behaviour in its next transaction. By evaluating the reputation values concerning previous transactions, the reputation system derives the probability that an agent will behave good or bad in the next transaction.

Adaptive Systems

It is a reputation system based on the following principles:

- The reputation value of a user is never lower than the reputation of a new user.

- Users with very high reputation values experience smaller rating changes than users with a low reputation.

- When two users interact more than once, only the most recent rating among them is considered.

ReGreT System

It is a reputation system in which agents themselves evaluate reputation in a decentralized way. In ReGreT, each agent has the capability to evaluate the reputation of other agents.

Perseus System

It is a personalized reputation system that aims to overcome the problems existing in most centralized systems for online marketplaces.