Coalition formation (CF) is frequently used as an efficient solution for achieving goals through resource sharing and exchange. However, most existing coalition formation methods suppose a global view of the system or the existence of facilitators which manage this view. This assumption is clearly unsuitable for wide multi-agent systems. In this paper, we propose a scalable coalition formation method adapted for large-scale distributed systems. In our approach, every agent has a partial view of the system and can only communicate with the agents of its own view.