What can be Learned from the Systematic Analysis of the Log Cluster of Agent Simulation

Abstract

In this paper, we propose a method to decompose a set of full logs into log clusters based on an organizational structure simulation model, clarify the properties of the log clusters, compare them and analyze them. This method is unique, in a sense that it analyzes at an intermediate level between the macro statistical analysis of the entire log and the micro level description analysis of one log. In this paper, we applied our proposed method to the result of existing research of organizational simulation. We found that log clusters containing known properties and unknown properties are consistent with the results of log analysis from existing studies. From the decision tree analysis, we also found a characteristic difference between the log clusters about the nature of the decision included within.