A clustering process for nodes in a network is required to find communities from social networks. General clustering algorithm needs to be configured the number of communities in advance. The number of communities is a very important element because the result of clustering can be different, depending on it. In this paper, we define the external community relationship which is distinguished between communities. Using the external community relationship as an evaluation metric of clustering result, we propose a method to determine the number of communities dynamically. We compare the proposed method to existing methods based on the accuracy of the number of communities and the average purity of communities. Our results show favorable performance for these criteria compared to the existing methods that were evaluated.