Self-Attentive Sequential Recommendation

Short summary

A new self-attention based sequential recommendation model (SASRec) captures long-term user behavior semantics while predicting the next item based on a few relevant historical actions, outperforming state-of-the-art models on both sparse and dense datasets and achieving an order of magnitude greater efficiency than comparable RNN/CNN models.

AI-generated from the title and abstract; the full text is not read.