In order to improve the accuracy of online education recommendation service on Hadoop platform and provide more accurate resource recommendation for different users, a high precision recommendation service strategy is proposed in this paper. Firstly, the Hadoop platform users were classified by weight estimation, then the online learning resource was labeled and classified, and the parameters of user interest resource was estimated. Finally, the recommendation strategy is generated. Experiments showed that, compared with ItemBased CF strategy and BehaviorBased CF strategy, the recommendation strategy had higher accuracy.