search

AaronHeee / Neural-Attentive-Item-Similarity-Model

TensorFlow Implementation of Neural Attentive Item Similarity Model for Recommendation on TKDE 2018
149 stars 48 forks source link

Can I ask a question about your code? #8

Closed SeongJinAhn closed 5 years ago

SeongJinAhn commented 5 years ago

https://github.com/AaronHeee/Neural-Attentive-Item-Similarity-Model/blob/751bb9ae3542f7dac43990b1772d38b07c6cb718/NAIS.py#L130

I think that, A means a_ij Hence, L130 means a_ij * qj.

image

https://github.com/AaronHeee/Neural-Attentive-Item-Similarity-Model/blob/751bb9ae3542f7dac43990b1772d38b07c6cb718/NAIS.py#L146

But when it comes to y_hat_ui (self.output), they multiply self.embedding_p (a_ij * q_j) and self.embedding_q again. I think it multiplies 'q_j' again.

Is there any reason why you multiply it again?

SeongJinAhn commented 5 years ago

I understand it thanks