Table 4 Dataset characteristics for name extractions
From: Network-theoretic information extraction quality assessment in the human trafficking domain
Metric | Set1 | Set2 | Set3 | Set4 | Set5 | Set6 | Set7 | Set8 |
|---|---|---|---|---|---|---|---|---|
No. of unique extractions | 925 | 631 | 900 | 468 | 594 | 1,076 | 1,218 | 1,169 |
No. of extractions per advertisement | 0.7563 | 0.6271 | 2.7060 | 0.3794 | 0.5140 | 1.0039 | 2.9484 | 0.6269 |
No. of ads with no extractions | 4,708 | 5,937 | 532 | 7,607 | 6,536 | 3,613 | 328 | 6,362 |
No. of ads with at least 1 extraction | 6,822 | 5,593 | 10,998 | 3,923 | 4,994 | 7,917 | 11,202 | 5,168 |
Precision | 1.0 | 0.6050 | 0.1899 | 1.0 | 1.0 | 0.7533 | 0.2565 | 0.6050 |
Recall | 1.0 | 0.5016 | 0.6796 | 0.5016 | 0.6796 | 1.0 | 1.0 | 0.5015 |
F-score | 1.0 | 0.5485 | 0.2969 | 0.6681 | 0.8092 | 0.8593 | 0.4083 | 0.5484 |