In traditional models of coalitional game theory, the outcome of a game is either the formation of the grand coalition, or partition of the set of players into disjoint coalitions (a coalition structure). However, if coalitions are associated with tasks, it is often possible for an agent to be involved in multiple tasks, and thus participate in several (overlapping) coalitions. In this work, we propose a model for overlapping coalition formation, and several notions of the core for the resulting class of games. We explore the relationships between these notions of the core, and analyze their computational complexity in a simple task-oriented domain.
Based on joint work with Georgios Chalkiadakis, Maria Polukarov, Evangelos Markakis, Nick Jennings and Yair Zick.
Back to Algorithmic Game Theory