Searching for Collaborators in Agent Networks

Research on social networks have received increasing attention in multiagent systems due to the popularity of peer-to-peer networks, social networking sites, electronic supply chains, etc. The common underlying research issue in these domains is the search and location of desirable interaction or collaboration partners in a relatively large population. This paper is motivated by some recent, intriguing research results involving agent-organized networks (AONs). In AONs agents have a limited number of collaboration partners at any time, represented by edges in a network of agent nodes, and can rewire edges, i.e., change partners, to improve performance. It was found that random selection of partners at every time period produced better performance but incurred larger search costs in a production and exchange economy compared to gradual rewiring of edges in the network. We propose an exponentially decaying exploration scheme that produces similar utilities to random rewiring but with much less rewiring costs. We also experiment with more realistic problem domains involving manufacturing profits, inventory costs, and stockpiling limits and observe superior utilities with our proposed searching scheme for collaborators.