Semantic-based dynamic positioning mechanism for problem solving in multi-agent systems

In multi-agent systems (MAS), finding agents which are able to service properly in an open and dynamic environment are the key issue in problem solving. However, it is difficult to find agent resources quickly and position agents accurately and complete the system integration by the keyword matching method, due to the lack of clear semantic information of the classical agent model. An semantic-based agent dynamic positioning mechanism was proposed to assist in the system dynamic integration. According to the semantic agent model and the description method, a two-stage process including the domain positioning stage and the service semantic matching positioning stage, was discussed. With this mechanism, proper agents that provide appropriate service to assign sub-tasks for task completion can be found quickly and accurately. Finally, the effectiveness of the positioning mechanism was validated through the in-depth performance analysis in the application of simulation experiments to the system dynamic integration.