Identification of the users' service preferences and discovery of user groups with similar preferences will have great significance for the development of personalized services recommendation system as well as user group-oriented customized services. This paper presented a concept of cloud community and designed a cloud community discovery algorithm based on group preferences. The algorithm mines service preferences similarity between cloud users based on a bipartite network structure, and then uses the improved agglomerative hierarchical clustering algorithm to identify cloud community structure. In the clustering process, group preferences intensity is introduced to determine the optimal community structure. Finally, the effectiveness of the algorithm was validated by the simulations.