Extracting social laws from unilateral binary constraint relation topologies in multiagent systems

摘要

Interaction relations are very important for the multiagents; and each agent will adopt different strategies in the interactions. There may be some social constraints in the system which can control the interaction relations and strategies among agents. A constraint may affect any number of agents from 1 to all, therefore, the impact domain of social constraint is local and different social constraints may also produce conflicts themselves. To receive the global harmony of all agents, some social laws should be endowed on the whole system which are the set of restrictions for agents’ strategies and are obeyed by all agents. This paper provides a model for extracting social laws from social constraints of agents. In the presented model, we mainly address the social law extraction from unilateral binary constraints. With the model, if the constraint relation topology (CRT) is a directed acyclic graph (DAG), then the extracted social law is unique; if the constraint relation topology (CRT) is a directed cyclic graph, then the extracted social law has the maximum usefulness degree.