Recommender systems have been viewed as powerful tools to filter overloaded information in the e - commerce environment. But traditional two - dimensional recommendation method s , which only explore the relevance between customers and products, are not applicable for the recommendation space in C2C (Customer to Customer) e - commerce context that involv es three types of entities : buyers, sellers and products. I n this paper, we propose a three - dimensional approach to explore the relevance among buyers, sellers and products, and provide personalized “seller and product” recommendations for buyers. First ly , similarities between sellers are calculated based on selle r features. Then the spare data in the three - dimensional historical rating set are supplemented and based on which buyer similarities are calculated to find neighbors who have similar product preferences with the target buyer. Finally, a three - dimensional rating prediction model is used to predict the unknown ratings that the buyer may give to candidate “seller and product” combinations. A real data experiment is conducted and the results prove the effectiveness of the proposed approach.