表 4. Comparison of recent approaches and our proposed method in cold relevance data set.
本文方法与新近方法在冷关联数据集上的实验结果对比
| 算法 | 数据集 | Accu | Rec | RMSE |
| user-based 方法 | MATADOR-C | 23.38% | 20.56% | 2.358 |
| STITCH-C | 22.78% | 19.78% | 2.669 | |
| item-based 方法 | MATADOR-C | 22.89% | 21.65% | 2.610 |
| STITCH-C | 20.36% | 19.35% | 2.721 | |
| Naïve Bayes 方法 | MATADOR-C | 25.45% | 24.36% | 2.221 |
| STITCH-C | 19.96% | 20.39% | 2.389 | |
| SVM 方法 | MATADOR-C | 25.92% | 26.75% | 2.003 |
| STITCH-C | 20.17% | 25.96% | 2.169 | |
| MTOI 方法 | MATADOR-C | 25.65% | 24.98% | 1.985 |
| STITCH-C | 24.89% | 23.54% | 2.069 | |
| 基于二分图的
核回归方法 |
MATADOR-C | 27.11% | 26.52% | 1.845 |
| STITCH-C | 26.05% | 24.85% | 1.896 | |
| 基于化学相似度
的方法 |
MATADOR-C | 26.25% | 25.59% | 1.956 |
| STITCH-C | 24.47% | 22.84% | 2.056 | |
| 本文方法 | MATADOR-C | 29.02% | 26.87% | 1.782 |
| STITCH-C | 27.54% | 24.87% | 1.806 |