We propose a novel proactive caching strategy with social community-aware learning in the Coordinated Multi-Point (CoMP)-enabled small-cell network. First, it predicts content request probabilities using community-aware non-negative matrix factorization (CNMF) with implicit feedback. The CNMF-based proactive caching algorithm then makes a caching decision by considering estimated content request probabilities of both the associated and neighbor users to closely cooperate with CoMP. Simulation results show that the proposed caching strategy improves the performance in terms of the content hit ratio and the average content delivery completion time according to the cache size. Furthermore, the performance of the proposed caching with and without CoMP is evaluated by varying the cache size and the small-cell base station density.