A Trust Model for Informed Agent Collaboration in Complex Tasks

The execution of a complex task, by definition, presumes that it cannot be performed by a single entity and requires collaboration and cooperation between two, or more, entities. Such collaboration necessities include the correct identification of trustworthy entities to engage with, and the possibility of maintaining a collaboration scenario (even with a market competitor) without given up sensible information that would prevent an organisation to sustain its competitiveness. And even after selecting the right partners, agents need to trust that the negotiation outcomes will be enforced. In this work we propose a model that enables the collaboration between entities to solve a complex task, while negotiating their preferences, with trust mechanisms associated with the reputation of each agent, and the enforcement of the negotiation outcomes. To achieve this goal, our model relies on the definition of a competitive multi-agent coalition network, and a reputation layer that allows them to perform informed decisions when engaging with other agents. This model is supported by an immutable and distributed structure, that uses blockchain as a layer for knowledge representation, and smart contracts as enforcers of a negotiation process.