The performance of idiom identification in a cross-domain setting where models are trained on MAGPIE random and tested on target domains (Tgt. Domain) SemEval5B random and VNC random. The performance is measured by Classification F1 score (F1;%) and Sequence Accuracy (SA;%).
Tgt. Domain . | Models . | F1 . | SA . |
---|---|---|---|
SemEval5B | RNN-MHCA | 81.35 | 54.72 |
DISC | 77.70 | 61.80 | |
VNC | RNN-MHCA | 85.01 | 69.74 |
DISC | 83.57 | 72.55 |
Tgt. Domain . | Models . | F1 . | SA . |
---|---|---|---|
SemEval5B | RNN-MHCA | 81.35 | 54.72 |
DISC | 77.70 | 61.80 | |
VNC | RNN-MHCA | 85.01 | 69.74 |
DISC | 83.57 | 72.55 |