源域数据增强与多兴趣细化迁移的跨域推荐模型

尹雅博,朱小飞,刘议丹

Cross-domain recommendation model based on source domain data augmentation and multi-interest refinement transfer

Yabo YIN,Xiaofei ZHU,Yidan LIU

表 3 CDR-ART在3个任务上的消融实验

Tab.3 Ablation experiment of CDR-ART on three tasks

任务 $ \beta $/% MAE CDR-ART w/o aux w/o refine w/o aux&&refine 任务1 20 0.812 3 0.847 0 0.865 4 0.943 5 50 0.958 3 1.048 7 1.041 3 1.084 3 80 1.176 7 1.255 7 1.255 9 1.342 9 任务2 20 0.874 4 0.887 1 0.897 2 0.920 2 50 0.922 7 0.935 7 0.938 0 0.961 8 80 0.997 2 1.027 4 1.026 4 1.044 3 任务3 20 0.869 7 0.909 9 0.944 5 0.972 8 50 0.978 7 1.040 8 1.038 7 1.080 9 80 1.131 2 1.189 7 1.195 8 1.245 9