Coalition formation is concerned with the question of how to partition a set of agents into disjoint coalitions according to their preferences. Deviating from most of the previous work, we consider an online variant of the problem, where agents arrive in sequence and whenever an agent arrives, they have to be assigned to a coalition immediately and irrevocably. The scarce existing literature on online coalition formation has focused on the objective of maximizing social welfare, a demanding requirement, even in the offline setting. Instead, we seek to achieve stable coalition structures in an online setting, and focus on stability concepts based on deviations by single agents. We present a comprehensive picture in additively separable hedonic games, leading to dichotomies, where positive results are obtained by deterministic algorithms and negative results even hold for randomized algorithms.

翻译：联盟形成问题关注如何根据代理人的偏好将一组代理人划分为不相交的联盟。与以往大多数研究不同，我们考虑该问题的一个在线变体，其中代理人按顺序到达，且每当一个代理人到达时，必须立即且不可撤销地将其分配至某个联盟。现有关于在线联盟形成的少量文献聚焦于最大化社会福利这一目标，而即便在离线场景下，该要求也十分苛刻。为此，我们致力于在在线环境中实现稳定的联盟结构，并重点关注基于单一代理人偏离行为的稳定性概念。我们在可加可分离享乐博弈中呈现了一个完整的图像，形成了二分法：肯定性结果可由确定性算法获得，而否定性结论甚至对随机化算法同样成立。