结合兴趣点类别周期属性和用户短期偏好特征的推荐模型

桑春艳, 易星宇, 廖世根, 文俊浩

A recommendation model combining point of interest category periodic attributes and user short⁃term preference features

Chunyan Sang, Xingyu Yi, Shigen Liao, Junhao Wen

表2 CPSTIN与几种基准模型在NYC和TKY数据集上的对比实验结果

Table 2 Experimental results of CPSTIN and benchmark models on the NYC and TKY datasets

NYC TKY Recall@5 Recall@10 MRR@5 MRR@10 Recall@5 Recall@10 MRR@5 MRR@10 OURS 0.6996 0.7491 0.5643 0.5710 0.6435 0.7065 0.5053 0.5138 NSSR 0.2446 0.2566 0.2162 0.2179 0.1936 0.2123 0.1677 0.1702 FPMC 0.3727 0.5392 0.2192 0.2377 0.1528 0.2155 0.0967 0.1049 ATSTLSTM 0.2923 0.3951 0.1859 0.1996 0.3165 0.4330 0.1894 0.2050 PLSPL 0.3221 0.3962 0.2117 0.2218 0.3438 0.4207 0.2271 0.2374 STAN 0.3634 0.5146 0.1988 0.2189 0.2500 0.3300 0.1645 0.1756 CORE 0.5226 0.5991 0.3738 0.3833 0.4208 0.4771 0.2975 0.3053 LSTM 0.5024 0.5986 0.3311 0.3437 0.4219 0.5108 0.2824 0.2944 LightSANs 0.5125 0.5771 0.3511 0.3600 0.4566 0.5556 0.3020 0.3155 HST⁃LSTM 0.5304 0.6230 0.3679 0.3807 0.4105 0.4873 0.2906 0.3009 GeoSAN 0.4297 0.5841 0.2592 0.2794 0.5629 0.6942 0.3626 0.3798 DeepMove 0.5782 0.6857 0.3847 0.3994 0.5032 0.6004 0.3358 0.3489 LSTPM 0.5993 0.7084 0.4184 0.4335 0.5031 0.5877 0.3564 0.3679