Table 2 The average of the weighted F1-scores across the five folds. On an average, there are 6030 training instances and 1507 test instances
From: Synthetic data for annotation and extraction of family history information from clinical text
Including “O” | Excluding “O” | |||||
|---|---|---|---|---|---|---|
System | Precision | Recall | F1-score | Precision | Recall | F1-score |
Dictionary baseline | 0.721 | 0.624 | 0.638 | 0.558 | 0.766 | 0.629 |
SVM | 0.843 | 0.843 | 0.841 | 0.781 | 0.738 | 0.756 |
CRF | 0.831 | 0.816 | 0.817 | 0.704 | 0.76 | 0.719 |