A Semantic-based Multi-agent Dynamic Interaction Model

Due to the autonomy of agents and their ability to perceive the environment, multi-agent systems have been widely used in many fields. The design of multi-agent systems requires the support of interactive models. The traditional multi-agent interaction model has certain feasibility in solving specific tasks. However, in a distributed environment, a relatively static multi-agent interaction model is not sufficient to support a dynamically changing interaction process. Frequent data interactions will also consume resources of multi-agent systems, thereby reducing agent performance. In this study, we propose a semantic-based multiagent dynamic interaction model (MADIM). MADIM uses semantic ontology to map the objects in the interaction model, and defines the interaction protocol through the rule description language. This model is attached with dynamically configurable semantic templates and interaction rule base. We added a reusable dynamic resolution engine component to MADIM to provide dynamic resolution services for the semantic information in the model. MADIM supports dynamic interactive behavior and has good interoperability and interpretability. Our model provides a flexible solution to the multi-agent interaction process. Finally, we verified the feasibility of the model design scheme through a simple example.