Detection of implied scenarios in multiagent systems with clustering agents' communications

Software agents in Multiagent Systems (MAS) have several interactions that are designed and represented in the scenarios of the system. These communications should be verified to detect whether the agents will show a new behavior in their execution, which is known as emergent behavior or implied scenario. Most research use different versions of state machines modeling for the detection of implied scenarios, which consider the states of one/all agents. The existing detection processes ignore the interactions among agents. In this paper, besides modeling the states and agents' behaviors, we model the agents' interactions derived from their designs, to detect implied scenarios. A new type of implied scenario that occurs when a process misses the information about its common communications in multiple scenarios is studied in this paper. This type of implied scenario cannot be detected with other approaches. Various situations that can lead to this implied scenario are ruled. Moreover, a detection methodology based on clustering the agents' communications from the scenarios of the system is presented. The results are verified through a case study.