TRAVOS: Trust and Reputation in the Context of Inaccurate Information Sources

Abstract

In many dynamic open systems, agents have to interact with one another to achieve their goals. Here, agents may be self-interested, and when trusted to perform an action for another, may betray that trust by not performing the action as required. In addition, due to the size of such systems, agents will often interact with other agents with which they have little or no past experience. There is therefore a need to develop a model of trust and reputation that will ensure good interactions among software agents in large scale open systems. Against this background, we have developed TRAVOS (Trust and Reputation model for Agent-based Virtual OrganisationS) which models an agent's trust in an interaction partner. Specifically, trust is calculated using probability theory taking account of past interactions between agents, and when there is a lack of personal experience between agents, the model draws upon reputation information gathered from third parties. In this latter case, we pay particular attention to handling the possibility that reputation information may be inaccurate.

CRM: An efficient trust and reputation model for agent computing

“…We also compare our model with five well known models as benchmarks: FIRE [10,11], Referral [33,34], SPORAS [35], Travos [27] and BRS [12]. All these models are explained in details and discussed in the related work section (Section 6).…”

“…Comparison is done first between CRM, FIRE [10] (a successful trust model with high performance), SPORAS [35] (a centralized approach), and Referral [33] (following the concept of reference in an honest environment). Travos [27] and BRS [12] are the two other models that we compared CRM with in terms of how they survive in biased environment where agents constantly change their behaviors. Like CRM, Travos and BRS are designed to take actions while agents are not fully trustworthy.…”

“…The simulation results outlined in this section are all based on 50% agent activation rate. Travos [27] has a method similar to BRS. It also uses beta distribution to estimate the trustworthiness of an agent based on the previous interaction experience.…”

“…According to [16], there are 3 main categories of trust evaluation models: predictive models [21,20,25], transactional models [12,20,27] and, mechanism design based models [13,36]. Predictive models have a sub-category of models using witness's information in order to increase accuracy of their trust measurement.…”

“…The most recent research proposals in trust models for multi-agent systems are as follows: (a) interaction trust, based on the direct interactions of two parties [30,31]; (b) trust based on the type of prior interactions [10,11,22]; (c) witness reputation based on certified (and possibly encrypted) references obtained by the agent to be evaluated after interacting with other agents. These references are then made public to any other agent who wants to interact with this agent [10,12,17,21,27]; and (d) referenced reputation, based on references from other agents detailing a particular agent's behavior [5,6,10,11].…”