EMSS 2012 Proceeding

Analysis of agents' behavior in multiagent system

Authors:   Katerina Slaninova, Jan Martinovic, Pavla Drazdilova, Dominik Vymetal, Roman Sperka

Abstract

Multi-agent systems are commonly used for simulation purposes. The authors focused on agent-based technology in the business process simulation, especially on the analysis of the agent-based simulation outputs in order to facilitate the verification of the used methodology. The paper deals with an analysis of agents? behavior in multi-agent model of business processes. The main goal of the paper is to find, how selected methods can influence finding of behavioral patterns of selected agents in the system and how the amount of extracted sequences can be reduced. Extraction of behavioral patterns was performed by process mining and pattern mining methods with the focus on the sequences. The authors present the comparison of selected methods for the definition of agents? behavior with the focus to selected characteristics of observed methods. Behavioral patterns and relations between them create complex networks; thus, the extraction of behavioral patterns is optimized by spectral graph partitioning. Moreover, the visualization of groups of agents with similar behavior is presented.

I3M  Scientific Sponsors

I3M  Industrial Sponsors

I3M  Media Sponsors