Efficient service-oriented inter-enterprise collaboration focuses on the business processes of the enterprises and hides the technology differences between them. However, such collaboration induces two main challenges. First, there should be an accessible and functioning infrastructure available for the collaboration. Second, as the growing number of participants can lead to the growing level of misbehaving among them, and thus market deterioration, that is why the parties should have common understanding of the behavior that is appropriate and can be trusted.
We focus on the trust relationships between the agents' interactions. Agents make a trust decision before interacting, and this decision among other factors is based on an estimation of another agent's reputation. A reputation management system collects and analyzes interactions experience between agents.
Let us call a person, company or any other possible entity whose goal is to build an infrastructure for the interactions between enterprises using one of the trust and reputation management algorithms as an infrastructure builder. For an infrastructure builder it is important to evaluate and compare reputation management systems, choosing one of them based on the evaluation and comparison results. This remains an open question in research. The thesis aims at supporting the decision-making process of this kind.
We suggest evaluation criteria for a trust or reputation management systems' evaluation. We implement a generic tool which can plug in a trust or reputation management algorithm and simulate the behavior of the multiagent system where every agent follows the same algorithm. We illustrate the tool's support for some of the suggested evaluation criteria. We provide some recommendations for further development of the generic tool for evaluating and comparing the above-mentioned behavior characteristics of different trust and reputation management algorithms.
ACM Computing Classification System (CCS):
Human-centered computing → Collaborative and social computing → Collaborative and social computing systems and tools → Reputation systems
Computing methodologies → Artificial intelligence → Distributed artificial intelligence → Multi-agent systems