Dependency Parsing on CoNLL German 2009 (test)
94.77UASSTACKPTR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| STACKPTRbeam size=102018.05 | 94.77 | 93.21 | |
| BIAF2018.05 | 94.52 | 93.06 | |
| BiAFModel Type=Graph-based, Variant=re-impl2018.05 | 93.85 | 92.32 | |
| STACKPTRModel Type=Transition-based, Variant=+sib2018.05 | 93.76 | 92.21 | |
| STACKPTRModel Type=Transition-based, Variant=+gpar2018.05 | 93.65 | 92.12 | |
| STACKPTRModel Type=Transition-based, Variant=Full2018.05 | 93.65 | 92.11 | |
| NeuroMST2018.05 | 93.62 | 91.9 | |
| STACKPTRModel Type=Transition-based, Variant=Org2018.05 | 93.59 | 92.06 | |
| BiAFModel Type=Graph-based, Variant=Original2018.05 | 93.46 | 91.44 | |
| Deep Biaffine2016.11 | 93.46 | 91.44 | |
| Bi-Att2018.05 | 92.71 | 89.8 | |
| Best Published2018.05 | 92.71 | 89.8 | |
| Ma and HovyModel Type=Graph-based2018.05 | 92.58 | 90.54 | |
| Kuncoro et al.Model Type=Graph-based2018.05 | 91.6 | 89.24 | |
| Bohnet and NivreModel Type=Transition-based2018.05 | 91.4 | 89.4 | |
| A'16-beamBeam size=>12016.03 | 90.91 | 89.15 | |
| Andor et al.Model Type=Transition-based2018.05 | 90.91 | 89.15 | |
| Andor et al.2016.11 | 90.91 | 89.15 | |
| Arc-hybrid, dyn. + PPTransition system=Arc-hybrid, Oracle=Dynamic (α=0.75), Pseudo-projective (PP)=true, Pre-training=true2016.03 | 90.34 | 88.17 | |
| Arc-hybrid, dyn. + PPTransition system=Arc-hybrid, Oracle=Dynamic (α=0.75), Pseudo-projective (PP)=true, Pre-training=false2016.03 | 89.8 | 87.29 | |
| Y'15Beam size=12016.03 | 89.6 | 86 | |
| A'16 + pre-trainingPre-training=true, Beam size=12016.03 | 89.12 | 86.95 | |
| Ballesteros et al.Model Type=Transition-based2018.05 | 88.83 | 86.1 | |
| Arc-standard, static + PPTransition system=Arc-standard, Oracle=Static, Pseudo-projective (PP)=true, Pre-training=true2016.03 | 88.56 | 86.15 | |
| Arc-standard, static + PPTransition system=Arc-standard, Oracle=Static, Pseudo-projective (PP)=true, Pre-training=false2016.03 | 88.09 | 85.24 |