Table 5.
Results on medical development set
| Strategy |
Abbr →Exp |
Exp →Abbr |
Syn |
||||||
|---|---|---|---|---|---|---|---|---|---|
| RI | RP | Result | RI | RP | Result | RI | RP | Result | |
|
RI ⊂ RP30 |
RI_4 |
RP_4_sw |
0.08 |
RI_2 |
RP_2 |
0.03 |
RI_20 |
RP_4_sw |
0.26 |
| RI_20 |
RP_2 |
RI_4 |
RP_4 |
||||||
| RI_20 |
RP_4_sw |
RI_4 |
RP_4_sw |
||||||
| |
|
RI_8 |
RP_8 |
||||||
| |
|
RI_20 |
RP_2 |
||||||
| |
|
RI_20 |
RP_2_sw |
||||||
| |
|
RI_20 |
RP_4 |
||||||
| |
|
RI_20 |
RP_4_sw |
||||||
|
RP ⊂ RI30 |
RI_2 |
RP_2_sw |
0.08 |
RI_2 |
RP_2 |
0.03 |
RI_8 |
RP_8_sw |
0.24 |
| RI_4 |
RP_4 |
RI_2 |
RP_2_sw |
||||||
| RI_4 |
RP_4_sw |
RI_4 |
RP_4 |
||||||
| RI_8 |
RP_8 |
RI_4 |
RP_4_sw |
||||||
| RI_8 |
RP_8_sw |
RI_8 |
RP_8 |
||||||
| RI_20 |
RP_2_sw |
RI_8 |
RP_8_sw |
||||||
| RI_20 |
RP_4 |
RI_20 |
RP_2 |
||||||
| RI_20 |
RP_4_sw |
RI_20 |
RP_2_sw |
||||||
| |
|
RI_20 |
RP_4 |
||||||
| |
|
RI_20 |
RP_4_sw |
||||||
| RI + RP | RI_4 | RP_4_sw | 0.10 | RI_8 |
RP_8_sw |
0.08 | RI_20 | RP_2_sw | 0.30 |
| RI_20 | RP_4_sw | ||||||||
Results (recall, top ten) of the best configurations for each model and model combination on the three tasks. The configurations are described according to the following pattern: model_windowSize. For RP, sw means that stop words are retained in the model.