Research Article

UPTDNet: A User Preference Transfer and Drift Network for Cross-City Next POI Recommendation

Table 5

Comparison of different methods for the next POI recommendation on the Washington-Baltimore dataset.

MethodWashington⟶BaltimoreBaltimore⟶Washington
MRRRec@1Rec@5Rec@10MRRRec@1Rec@5Rec@10

PRME0.18130.11930.26720.32100.16320.10260.22820.2667
RNN0.12710.08990.16150.18350.09960.07110.12630.1474
Time-LSTM0.13450.09570.16070.23420.13000.10130.15320.1740
Caser0.20250.13780.28570.32100.14300.10000.18460.2154
AttRec0.20170.15130.25550.29410.13780.10260.17440.1897
DeepMove0.21970.15270.30150.32820.13190.08820.18240.2176
LSTPM0.20360.15430.26560.28910.14290.09580.20000.2225
GETNext0.23890.18180.25000.31820.18890.15190.20250.2278
PLSPL0.22640.16970.28570.32770.16580.10260.23850.2795
Flashback0.20170.13950.27160.31010.18610.13420.22900.2895
Flashback-R0.22630.17250.27890.31740.19950.15790.23950.2684
UPTDNet0.26290.20730.32660.35600.21950.17110.28160.3105

The bold values indicate the best model results. Italic values indicate the second best model results.