SeqRec:基于长期偏好和即时兴趣的序列推荐模型
张岩,郭斌,王倩茹,张靖,於志文

SeqRec: sequential-based recommendation model with long-term preference and instant interest
Yan ZHANG,Bin GUO,Qian-ru WANG,Jing ZHANG,Zhi-wen YU
表 2 SeqRec模型与对比模型的召回率和命中率结果对比
Tab.2 Comparison of recall rate and hit rate results between SeqRec model and baselines
数据集 评价指标 R1 R5 R10 H1 H5 H10
Baby Caser 0.008 3(0.002 7) 0.022 5(0.003 0) 0.038 8(0.002 4) 0.015 0(0.005 0) 0.056 25(0.005 6) 0.110 0(0.003 1)
DREAM 0.003 0(0.001 0) 0.004 3(0.000 9) 0.033 7(0.004 4) 0.015 0(0.005 0) 0.031 30(0.004 0) 0.083 75(0.009 4)
SeqRec 0.006 3(0.001 8) 0.044 9(0.008 4) 0.056 9(0.009) 0.023 1(0.004 9) 0.080 0(0.010 4) 0.146 2(0.015 4)
提升幅度 −24% 99% 47% 54% 42% 33%
Electronics Caser 0.001 3(0.000 5) 0.010 0(0.003 1) 0.025 0(0.005 1) 0.007 5(0.002 0) 0.040 8(0.001 2) 0.073 3(0.006 5)
DREAM 0.000 5(0.000 4) 0.011 5(0.001 5) 0.021 5(0.001 5) 0.005 0(0.002 0) 0.040 8(0.009 2) 0.076 6(0.011 2)
SeqRec 0.001 4(0.001 0) 0.019 7(0.007 0) 0.027 8(0.004 7) 0.010 0(0.003 1) 0.047 5(0.008 2) 0.090 0(0.013 5)
提升幅度 8% 71% 11% 33% 16% 17%