| Abstract [eng] |
Multi-agent systems are a rapidly developing area of research. Languages are used for understandability of agent analysis, modeling and computerization processes for analysts as soon as system developers. The advantages of visual modeling are standard notation, unified concepts, and intuitive use. There are no single language suitable multi-agent systems modeling. Several languages were analyzed: AUML, objecticly oriented Petri networks, Agent Markup Language (AML), Agent Description Language (ADL), Q scenario description language, Multi-Agent Modeling Language (MAML), SCR++ - A Language Design for Real - Time Multi-Agent Systems), SDLM, with regards to their possibilities to represent and execute multi-agent system processes. It is proposed to extend UML functional, static and communication models with AUML stereotypes in multi-agents process modeling with succeeding transformation to execution language. Methodology was created describing a way to design multi-agent systems with UML models extended with AUML. The exclusive feature in proposed way of modeling lies in joining created models and making possible to analyst prepare program code for system developer. In order to test the proposed methodology scenario for modeling was prepared and models were designed: three models using UML notation, and one model using AUML extension. For implementation of proposed method, UML CASE tool Magic Draw was extended with stereotypes required for the proposed method of modeling. User templates were created to ease modeling based by proposed scenario. |
