On the Application of Clustering Techniques to Support Debugging Large-Scale Multi-Agent Systems

This work analyses the problematic of debugging a multiagent system. It starts from the fact that MAS are a particular type of distributed systems in which the active entities are autonomous in the sense that behavior and knowledge of the whole system is distributed among agents. It situates the problem by firstly studying the classical approaches for conventional code debugging and also the techniques used in distributed systems in general. From this initial perspective, it tries to situate agent and multi-agent systems debugging. It finally proposes the use of conventional data mining tasks like clustering to, by summarising, help in debugging huge MAS.