The amount of information in the world is increasing far more quickly than our ability to process it. Recommender systems are used by e-commerce sites to suggest products to their customers and to provide consumers with information to help them determine which products to purchase. In this paper, in a higher level, we analyze current personalized recommendation from two respects, respectively recommendation task and criteria for recommendation. Based on these analyses, a complete taxonomy on recommendation is build. Through the taxonomy, research routine and challenge in the future are identified.