Trust-Aware Sequence Recommendation Based on Attention Mechanism
Yejia Zeng, Zehui Qu, Bo Zhou · 2020
Existing trust-aware recommendation methods has problems such as inability to optimally allocate attention and ignore time sensitivity. Therefore, in order to improve the accuracy of the Top-N recommendation, we fuse the trust-based method with the sequence-based method. An attention-based trust recommendation sequence recommendation model (ATRec) was proposed, ATRec uses a self-attention block to learn short-term user preferences based on user interaction sequences, assigns users' attention to different trusted users through an attention mechanism, and finally balances the two with parameter. Our model has been tested on real data sets, and the experimental results are better than the baseline. Owing to the parallelism of self-attention, our model is also superior to the deep learning-based method in the baseline in terms of training speed.