Information overload forces online consumers to spend time and efforts searching for the desired products. To save time and decrease search cost, many online shopping websites have developed recommender systems. In recent years, many researchers have endeavored to develop hybrid recommender systems to improve the original systems. However, still quite a few consumers are reluctant to make purchases on online shopping websites, which results from different types of perceived risks. The study believes that two objective guidelines, (a) to build a learning relationship with consumers and (b) to know how to interact with consumers, will increase the possibility to succeed in concluding an online transaction for a recommender system. Above all, the study adopts the ideas of the E-K-B Model and George Miller's information processing theory to establish the learning framework by which the study conducts a business value chain for achieving the above guidelines. Moreover, by using Seetoo's strategic matrix analysis, a business model for recommender systems is developed. Finally, the study selects one successful online music website to show the feasibility of the proposed business model.