Table 3 Performance of the CRF-based concept extraction system, compared to the dictionary-based baseline on 100 news articles
From: Discovering opinion leaders for medical topics using news articles
| Â | True Positives | False Negatives | False Positives | Recall | Precision | F-measure |
|---|---|---|---|---|---|---|
Machine learning system | ||||||
Exact Match | 54 | 9 | 5 | 85.7 | 91.5 | 88.5 |
Partial Match | 55 | 8 | 4 | 87.3 | 93.2 | 90.1 |
Dictionary Baseline | ||||||
Exact Match | 45 | 18 | 254 | 71.4 | 15.0 | 24.9 |
Partial Match | 58 | 5 | 241 | 92.1 | 19.4 | 32.0 |