Negative Constraint Recommendation on ML 1M
0.2076Recall@10DPR (SAS)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DPR (SAS)Backbone=SASRec2026.02 | 0.2076 | 0.1772 | — | — | |
| DPRBackbone=SASRec2026.02 | 0.2076 | 0.1772 | 0.225 | 0.3172 | |
| SASRec Filter + fBackbone=SASRec, Variant=Tag Filtering + Fine-tuning2026.02 | 0.2075 | 0.1789 | 0.2278 | 0.3197 | |
| DPR (GRU)Backbone=GRU2026.02 | 0.203 | 0.1741 | — | — | |
| DPRBackbone=GRU4Rec2026.02 | 0.203 | 0.1741 | 0.222 | 0.3124 | |
| SASRec FilterBackbone=SASRec, Variant=Tag Filtering2026.02 | 0.1954 | 0.1666 | 0.2147 | 0.3057 | |
| SASRec Filter + f + aBackbone=SASRec, Variant=Tag Filtering + Implicit Fine-tuning2026.02 | 0.1827 | 0.1599 | 0.2026 | 0.2804 | |
| GRU4Rec Filter + fBackbone=GRU4Rec, Variant=Tag Filtering + Fine-tuning2026.02 | 0.1792 | 0.1509 | 0.1978 | 0.2872 | |
| SASRecBackbone=SASRec, Variant=Fine-tuning (+f)2026.02 | 0.1742 | 0.1463 | 0.1916 | 0.2784 | |
| GRU4Rec FilterBackbone=GRU4Rec, Variant=Tag Filtering2026.02 | 0.1728 | 0.1443 | 0.1884 | 0.2742 | |
| GRU4Rec Filter + f + aBackbone=GRU4Rec, Variant=Tag Filtering + Implicit Fine-tuning2026.02 | 0.169 | 0.1465 | 0.1886 | 0.2653 | |
| SASRecBackbone=SASRec, Variant=Base2026.02 | 0.1641 | 0.1363 | 0.1803 | 0.2653 | |
| SASRecBackbone=SASRec, Variant=Implicit Fine-tuning (+f+a)2026.02 | 0.1618 | 0.1382 | 0.1772 | 0.2507 | |
| GRU4RecBackbone=GRU4Rec, Variant=Implicit Fine-tuning (+f+a)2026.02 | 0.1499 | 0.1281 | 0.1675 | 0.2403 | |
| GRU4RecBackbone=GRU4Rec, Variant=Fine-tuning (+f)2026.02 | 0.1497 | 0.123 | 0.1651 | 0.2471 | |
| GRU4RecBackbone=GRU4Rec, Variant=Base2026.02 | 0.1432 | 0.1171 | 0.1579 | 0.2375 | |
| RecGPT-7BEvaluation Protocol=Fine-tuned2026.02 | 0.1277 | 0.087 | — | — | |
| RecLM-genEvaluation Protocol=Fine-tuned2026.02 | 0.1163 | 0.0773 | — | — | |
| GPT-3.5Evaluation Protocol=Zero-shot, Evaluation subset=1,000 users2026.02 | 0.0379 | 0.0228 | — | — | |
| DeepSeek-V3Evaluation Protocol=Zero-shot, Evaluation subset=1,000 users2026.02 | 0.035 | 0.0201 | — | — | |
| Qwen2.5-7BEvaluation Protocol=Zero-shot2026.02 | 0.0166 | 0.013 | — | — | |
| Llama-2-7bEvaluation Protocol=Zero-shot2026.02 | 0.0103 | 0.0081 | — | — |