An Agent Model Based on Open Linked Data for Building Internet of Agents Ecosystems

This paper presents an smart, collaborative and self-adaptive reactive agent model aimed at managing the resources of objects connected to Internet of Things (IoT). This agent model, called Linked Open Agent (LOA), is described using both a semantic agent contract built from descriptors published as linked data, and a workflow for agent control that is completed at runtime by the agent itself to address its behavior. The accuracy for semantic discovering agents partners was evaluated and compared with generic models of discovery such as the Yellow Pages of Java Agent DEvelopment Framework (JADE) and the Java implementation of the Universal Description, Discovery, and Integration (jUDDI). The results demonstrated that our method had a better accuracy for recovering agents than the accuracy of JADE and JUDDI.