Emergence of Complex Networks through Local Optimization

Abstract. The emergence of complex network structures of relationships between autonomous agents occurs in a wide range of distributed systems. Many researchers have proposed models to explain and reproduce this phenomenon. However, the mechanisms proposed require implausible assumptions of global knowledge about structural positions, and do not make explicit individual agents’ goals that motivate decisions to establish or break ties. We propose a model grounded in social exchange theory that is based on a population of agents with dissimilar attractiveness levels, who seek to optimize the set of their exchange partners in terms of the outcome they receive in a support game. In our model, agents perform a local optimization process constrained by realistic and plausible assumptions of local and imperfect information. We show that our model can generate several types of complex networks, notably power-law, small-world and center-periphery networks. This diversity of network structures...

• Type: Conference
• Year: 2004
• Where: ECAI
• Authors: Josep M. Pujol, Andreas Flache, Ramon Sangüesa, Jordi Delgado
• Keywords: Artificial Intelligence, Autonomous Agents, Complex Network Structures, Complex Networks, ECAI 2004