AMR parsing on LDC2014T12 Newswire section
73.3F1 ScoreTransition-based AMR Parser
Evaluation Results
| Method | Links | |
|---|---|---|
| Transition-based AMR ParserInput features=Word + POS, Aligner=Our aligner, Ensemble size=102018.10 | 73.3 | |
| Transition-based AMR ParserInput features=Word only, Aligner=Our aligner, Ensemble size=102018.10 | 72.5 | |
| Transition-based AMR ParserInput features=Word + POS, Aligner=Our aligner, Ensemble size=32018.10 | 72.5 | |
| Transition-based AMR ParserInput features=Word only, Aligner=Our aligner, Ensemble size=32018.10 | 71.9 | |
| Zhou et al. (2016)POS=true, DEP=true, NER=true, SRL=true, Pipeline System=true2017.07 | 71 | |
| Zhou et al.2018.10 | 71 | |
| Wang et al. (2015a)POS=true, DEP=true, NER=true, SRL=true, Pipeline System=true2017.07 | 70 | |
| Wang et al.2018.10 | 70 | |
| Goodman et al.2018.10 | 70 | |
| Transition-based AMR ParserInput features=Word + POS, Aligner=Our aligner, Ensemble size=12018.10 | 69.8 | |
| Transition-based AMR ParserInput features=Word only, Aligner=Our aligner, Ensemble size=12018.10 | 69.3 | |
| Stack-LSTM AMR parserPre-trained word embeddings=true, Character-based representations=true, POS=true, DEP=true2017.07 | 69 | |
| Ballesteros and Al-OnaizanInput features=Word + POS + DEP2018.10 | 69 | |
| Transition-based AMR ParserInput features=Word + POS, Aligner=JAMR aligner, Ensemble size=12018.10 | 68.8 | |
| Transition-based AMR ParserInput features=Word only, Aligner=JAMR aligner, Ensemble size=12018.10 | 68.6 | |
| Stack-LSTM AMR parserPre-trained word embeddings=true, Character-based representations=true2017.07 | 68 | |
| Stack-LSTM AMR parserPre-trained word embeddings=true, Character-based representations=true, POS=true2017.07 | 68 | |
| Ballesteros and Al-OnaizanInput features=Word only2018.10 | 68 | |
| Ballesteros and Al-OnaizanInput features=Word + POS2018.10 | 68 | |
| Artzi et al.2018.10 | 66.3 | |
| Artzi et al. (2015)POS=true, CCG=true2017.07 | 66 | |
| Stack-LSTM AMR parserPre-trained word embeddings=false, Character-based representations=true2017.07 | 66 | |
| Stack-LSTM AMR parserPre-trained word embeddings=true, Character-based representations=false2017.07 | 66 | |
| Stack-LSTM AMR parserPre-trained word embeddings=false, Character-based representations=false2017.07 | 64 | |
| Wang et al. (2015b)POS=true, DEP=true, NER=true, Pipeline System=true2017.07 | 63 | |
| Werling et al. (2015)POS=true, DEP=true, NER=true, Pipeline System=true2017.07 | 62 | |
| Flanigan et al. (2014)POS=true, DEP=true, Pipeline System=true2017.07 | 59 |