Table 7.
Disjoint corpus ensemble results on clinical + medical development set
| Strategy | Normalize |
Abbr →Exp |
Exp →Abbr |
Syn |
|||
|---|---|---|---|---|---|---|---|
|
Clinical |
Medical |
Clinical |
Medical |
Clinical |
Medical |
||
|
RI_4
|
RI_4
|
RI_4
|
RI_8
|
RI_20
|
RI_20
|
||
| RP_4_sw | RP_4_sw | RP_4_sw | RP_8_sw | RP_4_sw | RP_2_sw | ||
| AVG |
True |
0.13 |
0.09 |
0.39 |
|||
| AVG |
False |
0.24 |
0.11 |
0.39 |
|||
| SUM |
True |
0.13 |
0.09 |
0.34 |
|||
| SUM |
False |
0.32 |
0.17 |
0.52 |
|||
| AVG →AVG |
|
0.15 |
0.09 |
0.41 |
|||
| SUM →SUM |
|
0.13 |
0.07 |
0.40 |
|||
| AVG →SUM |
|
0.15 |
0.09 |
0.41 |
|||
| SUM →AVG | 0.13 | 0.07 | 0.40 | ||||
Results (P = weighted precision, R = recall, top ten) of the best models with and without post-processing on the three tasks. Dynamic # of suggestions allows the model to suggest less than ten terms in order to improve precision. The results are based on the application of the model combinations to the development data.