Recommender systems (RS) are found in many online applications where users are exposed to huge sets of items. The goal of recommender systems is to provide the users with a list of recommended items that they prefer, or predict how much they might prefer each item. Collaborative Filtering (CF) is a commonly used technique in RS. This approach recommends user based on the preferences of other similar users. Nowadays, several e-commerce sites such as Last.fm, Delicious, Epinions allow users to build their own social networks in the systems. Customers are able to connect with others, share their comments and reviews, thus forming a social network. One common property observed in social networks is that they exhibit community structure. Several algorithms have been proposed to automatically discover these communities. One question is whether we can provide better recommendations based on the opinions from users communities. In this paper, we assess the effectiveness of community-based approach in CF task. We consider the communities discovered by different features: local, global, unipartite, bipartite, structural and attributed communities. Experimental results on several real-world datasets show that these methods bring certain improvements in CF.
