From: Improving chemical disease relation extraction with rich features and weakly labeled data
 | Features | Precision (%) | Recall (%) | F-value (%) | F-value change (%) |
---|---|---|---|---|---|
1 | All features | 64.24 | 52.06 | 57.51 | Â |
2 | - BOW | 63.09 | 51.31 | 56.60 | −0.91 |
3 | - BOB | 61.24 | 52.63 | 56.61 | −0.90 |
4 | - Pattern | 61.83 | 51.22 | 56.03 | −1.48 |
5 | - Shortest path | 62.03 | 52.72 | 57.00 | −0.51 |
6 | - Statistical | 53.29 | 41.74 | 46.82 | −10.69 |
7 | - #1 ~ #8 | 62.54 | 50.75 | 56.03 | −1.48 |
8 | - #1 and #2 | 62.90 | 51.69 | 56.75 | −0.76 |
9 | - #3 and #4 | 63.31 | 51.97 | 57.08 | −0.43 |
10 | - #5 ~ #8 | 63.23 | 51.78 | 56.94 | −0.57 |
11 | - #9 ~ #11 | 54.04 | 45.12 | 49.18 | −8.33 |
12 | - #9 | 63.62 | 52.16 | 57.32 | −0.19 |
13 | - #10 | 57.09 | 45.31 | 50.52 | −6.99 |
14 | - #11 | 61.49 | 50.47 | 55.44 | −2.07 |
15 | - #12 ~ #19 | 63.79 | 52.06 | 57.33 | −0.18 |