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 2 模型在3个任务上的总体表现

Tab.2 Overall performance of model on three tasks

任务 模型 MAE RMSE β=20% β=50% β=80% β=20% β=50% β=80% 任务1 TGT 4.480 3 4.498 9 4.502 0 5.158 0 5.173 6 5.189 1 CMF 1.520 9 1.689 3 2.418 6 2.015 8 2.227 1 3.093 6 DCDCSR 1.491 8 1.814 4 2.719 4 1.921 0 2.343 9 3.306 5 SSCDR 1.301 7 1.376 2 1.504 6 1.657 9 1.747 7 1.922 9 EMCDR 1.235 0 1.327 7 1.500 8 1.551 5 1.664 4 1.877 1 PTUPCDR 1.150 4 1.280 4 1.404 9 1.519 5 1.638 0 1.823 4 CDR-ART 0.812 3 0.958 3 1.176 7 1.160 0 1.337 2 1.545 7 任务2 TGT 4.183 1 4.228 8 4.212 3 4.753 6 4.792 0 4.814 9 CMF 1.363 2 1.581 3 2.157 7 1.791 8 2.088 6 2.677 7 DCDCSR 1.397 1 1.673 1 2.361 8 1.734 6 2.055 1 2.770 2 SSCDR 1.239 0 1.213 7 1.317 2 1.652 6 1.560 2 1.702 4 EMCDR 1.116 2 1.183 2 1.315 6 1.412 0 1.498 1 1.643 3 PTUPCDR 0.997 0 1.089 4 1.199 9 1.331 7 1.439 5 1.591 6 CDR-ART 0.874 4 0.922 7 0.997 2 1.159 1 1.241 2 1.322 6 任务3 TGT 4.487 3 4.507 3 4.620 4 5.167 2 5.172 7 5.230 8 CMF 1.828 4 2.128 2 3.013 0 1.382 9 2.727 5 3.694 8 DCDCSR 1.841 1 2.173 6 3.140 5 2.295 5 2.677 1 3.584 2 SSCDR 1.541 4 1.473 9 1.641 4 1.928 3 1.844 1 2.140 3 EMCDR 1.352 4 1.473 2 1.719 1 1.673 7 1.800 0 2.111 9 PTUPCDR 1.228 6 1.376 4 1.578 4 1.608 5 1.744 7 2.051 0 CDR-ART 0.869 7 0.978 7 1.131 2 1.244 3 1.324 0 1.515 5