Table 5.
Precision on the stratified sample. For each error category, a random list of 15 GeneRIFs were independently examined by the two judges.
| No. | Category | True Positive | False Positive | Precision |
|---|---|---|---|---|
| 1. | Discontinued | 15 | 0 | 100% |
| 2. | Misspellings | 15 | 0 | 100% |
| 3. | Punctuation | 13 | 2 | 86.7% |
| 4. | Computational methods | 15 | 0 | 100% |
| 5. | Similar GeneRIFs | 15 | 0 | 100% |
| 6. | One-to-many | 15 | 0 | 100% |
| 7. | Length constraint | 5 | 10 | 33.3% |
| 8. | Overall | 93 | 12 | 88.6% |