Mining Core Motivations among Motivational Agents

Motivation is an important factor in reasoning about rational behavior of intelligent agents and analyzing the property of social network circles. Recent study on motivational agent paid their main attention on the mechanism of reasoning and multi-agent Cooperation. How motivation affects the internal structure of the allied agent groups are less considered. This paper proposes a methodology for motivational agent clustering, cohesion property analyzing and core motivational agent identifying. The methodology first finds clustered agents from the underlying graph that captures the similarity based interconnection topology of the agents. Then, the subgroups of agents that have high degrees of connectivity are extracted which can be thought of as the key representatives of the whole agent clusters. Our empirical results on real survey data and simulation platform show that our method is quite favorable for clearly partitioning large body of motivational agents and helping the analyzer to identify internal structure of the agent groups. Our algorithms can be adapted in various ways for social network behavior analyzing, intrusion detection and marketplace bidding strategy designing.