| 1. |
Using fuzz and the title field, 581 CTRI matches were picked up for the 581 CTG records. The model predicted 434 of these to be true. |
434 |
74.7% |
| 2. |
Assuming that fuzz picked up 80% of the true matches, the number of correct matches would be 543. |
543 |
93.5% |
| 3. |
Incorrect matches, as predicted by fuzz. |
38 |
6.5% |
| 4. |
Using tfidf and the title field, 581 CTRI matches were picked up for the 581 CTG records. The model predicted 441 of these to be true. |
441 |
75.9% |
| 5. |
Assuming that tfidf picked up 81% of the true matches, the number of correct matches would be 544. |
544 |
93.6% |
| 6. |
Incorrect matches, as predicted by tfidf. |
37 |
6.4% |
| 7. |
CTRI matches picked up by both fuzz and tfidf. |
304 |
52.3% |
| 8. |
Of the 304 CTRI matches picked up by both fuzz and tfidf, the model predicted 288 to be true. |
288 |
49.6% |
| 9. |
Incorrect matches from the common cases. |
293 |
50.4% |
