Sequential Recommendation on Amazon Beauty (Leave-one-out)
37.1NDCG@10ABKD-2
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| ABKD-2Backbone=SASRec, alpha=0.5, beta=1.52026.06 | 37.1 | — | — | 44.1 | |
| BAHSDBackbone=SASRec2026.06 | 35.8 | — | — | 52.5 | |
| TeacherBackbone=SASRec2026.06 | 34.1 | — | — | 49.3 | |
| CDBCFBackbone=SASRec2026.06 | 33.7 | — | — | 48.8 | |
| ME-MIABackbone=SASRec2026.06 | 33.4 | — | — | 48.6 | |
| DFMEBackbone=SASRec2026.06 | 33.3 | — | — | 48.3 | |
| TeacherBackbone=BERT4Rec2026.06 | 32.1 | — | — | 49.3 | |
| BAHSDBackbone=BERT4Rec2026.06 | 29.7 | — | — | 46.5 | |
| ABKD-1Backbone=SASRec, alpha=1.2, beta=1.22026.06 | 26.3 | — | — | 43.2 | |
| ABKD-3Backbone=SASRec, alpha=1.5, beta=0.52026.06 | 26.3 | — | — | 43.5 | |
| ABKD-1Backbone=BERT4Rec, alpha=1.2, beta=1.22026.06 | 21.2 | — | — | 35.5 | |
| ME-MIABackbone=BERT4Rec2026.06 | 19 | — | — | 26.7 | |
| DFMEBackbone=BERT4Rec2026.06 | 18.8 | — | — | 26.5 | |
| ABKD-2Backbone=BERT4Rec, alpha=0.5, beta=1.52026.06 | 18.8 | — | — | 27.1 | |
| CDBCFBackbone=BERT4Rec2026.06 | 18.6 | — | — | 25.7 | |
| ABKD-3Backbone=BERT4Rec, alpha=1.5, beta=0.52026.06 | 18.1 | — | — | 26.6 | |
| UnKDBackbone=BERT4Rec2026.06 | 14.7 | — | — | 24.9 | |
| DHKDBackbone=BERT4Rec2026.06 | 10.9 | — | — | 21 | |
| UnKDBackbone=SASRec2026.06 | 10.5 | — | — | 20.6 | |
| DHKDBackbone=SASRec2026.06 | 6.9 | — | — | 13.7 | |
| SASRec-SCEData splitting strategy=Leave-one-out2024.09 | 5.44 | 9.35 | — | — | |
| CBiTData splitting strategy=Leave-one-out2024.09 | 5.37 | 9.05 | — | — | |
| SASRec-RECEData splitting strategy=Leave-one-out2024.09 | 5.25 | 8.97 | — | — | |
| SASRec + BFTBackbone=SASRec, BFT Integration=true2026.05 | 5.06 | 8.9 | 4.55 | — | |
| SASRecBackbone=SASRec, BFT Integration=false2026.05 | 5.04 | 8.81 | 4.53 | — | |
| BERT4Rec + BFTBackbone=BERT4Rec, Model variant=+BFT, Ranking protocol=full-ranking, Number of random seeds=202026.05 | 4.63 | 9.01 | 4.14 | — | |
| FEARecData splitting strategy=Leave-one-out2024.09 | 4.59 | 8.84 | — | — | |
| BERT4RecBackbone=BERT4Rec, Model variant=Base, Ranking protocol=full-ranking, Number of random seeds=202026.05 | 4.52 | 8.76 | 4.06 | — | |
| DuoRecData splitting strategy=Leave-one-out2024.09 | 4.43 | 8.45 | — | — | |
| CL4SRecData splitting strategy=Leave-one-out2024.09 | 2.99 | 6.81 | — | — |