Integrating indicators of trustworthiness into reputation-based trust models: Insurance, certification, and coalitions

Abstract

Reputation-based trust models are essentially reinforcement learning mechanisms reliant on feedback. As such, they face a cold start problem when attempting to assess an unknown service partner. State-of-the-art models address this by incorporating dispositional knowledge, the derivation of which is not described regularly. We propose three mechanisms for integrating knowledge readily available in cyber-physical services (e.g., online ordering) to determine the trust disposition of consumers towards unknown services (and their providers). These reputation-building indicators of trustworthiness can serve as cues for trust-based decision making in eCommerce scenarios and drive the evolution of reputation-based trust models towards trust management systems. © 2012 IFIP International Federation for Information Processing.