To solve the information overload problem of large scale E-commerce systems in the big data era, a solution based on Hadoop is proposed, aiming at building a distributed recommendation system. Data analysis algorithms based on MapReduce programming model have high scalability and good performance. To overcome the limit of single recommendation technology, a hybrid model is adopted, which combines several complementary methods. Empirical studies show that the recommendation system on Hadoop has good scalability and efficiency.
