Table 2 Accuracy for different experiment setting
From: Enhancing unsupervised medical entity linking with multi-instance learning
|  | E+ | E− | Train stage | Accuracy (%) |
|---|---|---|---|---|
BEL | – | – | – | 58.62 |
Baseline | Top1 | Random Top20 | Pre-train | 58.62 |
EMEL-top | Top3 | Random Top20 | MIL | 59.77 |
Top3 | Random all | MIL | 58.76 | |
Top4 | Random Top20 | MIL | 59.19 | |
Top4 | Random all | MIL | 57.90 | |
EMEL-cluster | Cluster2 | Random Top20 | MIL | 59.19 |
Cluster3 | Random Top20 | MIL | 60.34 | |
Cluster4 | Random Top20 | MIL | 60.05 |