Table 7. Performance with training datasets of various sizes using gold standard medical concepts.
Data Size | Explicit recognizer | Implicit recognizer | Micro | INC | ||||
P | R | F | P | R | F | F | ||
50 | 0.9316 | 0.8245 | 0.8748 | 0.9527 | 0.6364 | 0.7631 | 0.8491 | / |
100 | 0.9384 | 0.8483 | 0.8911 | 0.9593 | 0.6966 | 0.8071 | 0.8717 | 0.0226 |
150 | 0.9423 | 0.8621 | 0.9004 | 0.9654 | 0.7147 | 0.8213 | 0.8821 | 0.0330 |
200 | 0.9479 | 0.8713 | 0.9080 | 0.9664 | 0.7418 | 0.8393 | 0.8922 | 0.0431 |
250 | 0.9488 | 0.8743 | 0.9102 | 0.9678 | 0.7595 | 0.8511 | 0.8964 | 0.0473 |
300 | 0.9481 | 0.8776 | 0.9114 | 0.9686 | 0.7631 | 0.8537 | 0.8978 | 0.0487 |
*Inc is increment compared with baseline.