Algorithms for Transitive Dependence-Based Coalition Formation

Coalition formation methods allow autonomous agents to join together in order to act as a coherent group in which they increase their individual gains by collaborating with each other. Although there are some research efforts toward coalition formation in multiagent systems (MAS), such as game theory-based approaches, these methods cannot be easily applied in real-world scenarios. Based on a novel social reasoning theory, namely, transitive dependence theory, this work proposes two dynamic coalition formation algorithms for coalition formation: 1) without and-action dependence and 2) with and-action dependence, respectively. While most related work addresses the problem of searching for the optimal coalition structure (CS), the proposed algorithms aim to find out the optimal coalitions for specific goals. Theoretical analysis and experimental results suggest that 1) the algorithm for coalition formation without and-action dependence is of polynomial complexity and is efficient, and 2) when the incidence rate of and-action dependence is not high, the anytime algorithm for coalition formation with and-action dependence is also efficient although it has relatively high complexity (NP-complete).