Research Article

Session-Based Recommendation with GNN and Time-Aware Memory Network

Table 4

The performance of SR-GTM with different position embedding method (%).

MethodYoochoose1/64RetailRocket
P@10MRR@10P@20MRR@20P@10MRR@10P@20MRR@20

SR-GTM-NP61.0931.6671.6132.3955.8236.2463.1536.75
SR-GTM-FP61.0931.5071.7032.2456.1836.4863.4136.98
SR-GTM61.3031.7871.8232.5356.5736.9163.6437.41