Deep Interest-Shifting Network with Meta-Embeddings for Fresh Item Recommendation

Table 4

Testing AUC comparison of context-aware models on the Taobao-Fresh dataset.

Methods	Auxiliary	Context	Full data
DeepFM	0.7367	0.7449	0.7362
PNN	0.7413	0.7439	0.7417
CFM	0.7377	0.7441	0.7442
DisNet-Add	–	0.7480	0.7528
DisNet-COT	–	0.7467	0.7533
DisNet-NN	0.7409	0.7483	0.7534

indicates whether the RM-IdEG variant is significantly superior to the coupling algorithm or not (pairwise t-test at the 0.05 significance level).On the auxiliary-only data, the network architecture of DisNet is fixed, and we only report the performance once.