session-based recommendation focuses on the prediction of user actions based
on anonymous sessions and is a necessary method in the lack of user historical
data. However, none of the existing session-based recommendatio
本文提出了基于互补学习和课程学习的新框架 Mutual Enhancement of Long-Tailed user and item (MELT),它是序列推荐系统中第一篇同时解决长尾用户和物品的问题,同时也是模型无关的。实验结果表明,该方法可以有效地缓解长尾问题,并且不会牺牲头部用户和物品的性能。