zTrust: Adaptive Decentralized Trust Model for Quality of Service Selection in Electronic Marketplaces

摘要

We present an adaptive decentralized trust formalization well suited for electronic commerce. Our model, called zTrust, constitutes two essential elements. The first is the adviser modeling mechanism that enables consumer agents to merge the cognitive and the probabilistic views of trust and adaptively calculate the trustworthiness of advisers according to environmental conditions, information availability, and participants' behavioral dispositions. Using this mechanism, consumers are able to form their social network consisting of the most reliable advisers. The second element is a trust-oriented service selection framework that models the qualification and trustworthiness of providers in delivering the multiattribute products and adopts a procurement auction model to choose the most pertinent provider that meets a consumer's quality of service requirements. We give a formal description of our approach and validate it with simulations demonstrating that our solution yields high-quality results under various realistic conditions. Experimental results indicate that the zTrust model can be effectively employed in dynamic agent-oriented e-commerce applications.