Semantic Role Labeling on CoNLL Brown 2005 (test)
93.7F1LISA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| LISAEmbeddings=GloVe2018.04 | 93.7 | 95.5 | 91.9 | — | |
| He et al. (2017)2018.04 | 92.4 | 89.3 | 95.7 | — | |
| Span-based SRLGold predicates=true, ELMo=true2018.05 | 80.4 | — | — | — | |
| LISAEmbeddings=ELMo, Parses=D&M2018.04 | 78.25 | 79.02 | 77.49 | — | |
| LISAEmbeddings=ELMo, Parses=LISA predicted2018.04 | 78.05 | 78.95 | 77.17 | — | |
| LISAEmbeddings=GloVe, Predicates=Gold, Parser=D&M parses2018.04 | 76.54 | 76.65 | 76.44 | — | |
| SAEmbeddings=ELMo, Parses=None2018.04 | 76.35 | 77.1 | 75.61 | — | |
| Span-based SRL (Ours)Embeddings=ELMo, Evaluation Protocol=End-to-End2018.05 | 76.1 | 73.9 | 78.4 | — | |
| He et al. (2018)Embeddings=ELMo2018.04 | 76.1 | 73.9 | 78.4 | — | |
| DEEPATTArchitecture=FFN, Ensemble=true2017.12 | 74.8 | 74.6 | 75 | 48.6 | |
| Tan et al. (2018)Gold predicates=true, Product of Experts (POE)=true2018.05 | 74.8 | — | — | — | |
| LISAEmbeddings=GloVe, Parses=D&M2018.04 | 74.66 | 75.8 | 73.54 | — | |
| LISAEmbeddings=GloVe, Predicates=Gold, Parser=LISA parses2018.04 | 74.55 | 74.77 | 74.32 | — | |
| DEEPATTArchitecture=FFN, Ensemble=false2017.12 | 74.1 | 73.5 | 74.6 | 48.4 | |
| Tan et al. (2018)Gold predicates=true2018.05 | 74.1 | — | — | — | |
| Tan et al. (2018)Embeddings=GloVe, Predicates=Gold2018.04 | 74.1 | 73.5 | 74.6 | — | |
| Span-based SRLGold predicates=true, ELMo=false2018.05 | 73.7 | — | — | — | |
| He et al. (2018)Embeddings=GloVe, Predicates=Gold2018.04 | 73.7 | 74.2 | 73.1 | — | |
| He et al.Ensemble=true2017.12 | 73.6 | 74.9 | 72.4 | 46.5 | |
| He et al. (2017)Gold predicates=true, Product of Experts (POE)=true2018.05 | 73.6 | — | — | — | |
| SAEmbeddings=GloVe, Predicates=Gold, Parser=None (syntax-free)2018.04 | 73.13 | 73.89 | 72.39 | — | |
| DEEPATTArchitecture=RNN, Ensemble=false2017.12 | 72.9 | 72.5 | 73.4 | 44.7 | |
| DEEPATTArchitecture=CNN, Ensemble=false2017.12 | 72.7 | 72.8 | 72.7 | 45.9 | |
| FitzGerald et al.Architecture=Struct., Ensemble=true2017.12 | 72.2 | 74.5 | 70 | 41.3 | |
| FitzGerald et al. (2015)Gold predicates=true, Product of Experts (POE)=true2018.05 | 72.2 | — | — | — | |
| He et al.Ensemble=false2017.12 | 72.1 | 72.8 | 71.4 | 44.8 | |
| He et al. (2017)Gold predicates=true2018.05 | 72.1 | — | — | — | |
| Yang and Mitchell (2017)Gold predicates=true2018.05 | 72 | — | — | — | |
| LISAEmbeddings=GloVe, Parses=LISA predicted2018.04 | 71.91 | 73.32 | 70.56 | — | |
| SAEmbeddings=GloVe, Parses=None2018.04 | 71.51 | 72.98 | 70.1 | — | |
| Tackström et al.Architecture=Struct., Ensemble=false2017.12 | 71.3 | 74.3 | 68.6 | 39.8 | |
| Span-based SRL (Ours)Embeddings=GloVe, Evaluation Protocol=End-to-End2018.05 | 70.8 | 69.7 | 71.9 | — | |
| He et al. (2018)Embeddings=GloVe2018.04 | 70.8 | 69.7 | 71.9 | — | |
| He et al. (2017)Model Variant=Product of Experts (POE), Evaluation Protocol=End-to-End2018.05 | 70.1 | 69.7 | 70.5 | — | |
| He et al. (2017) PoEEmbeddings=GloVe2018.04 | 70.1 | 69.7 | 70.5 | — | |
| Zhou and Xu2017.12 | 69.4 | 70.7 | 68.2 | — | |
| Zhou and Xu (2015)Gold predicates=true2018.05 | 69.4 | — | — | — | |
| Toutanova et al.Ensemble=true2017.12 | 68.8 | — | — | 40.8 | |
| He et al. (2017)Evaluation Protocol=End-to-End2018.05 | 68.5 | 67.6 | 69.6 | — | |
| Punyakanok et al.Ensemble=true2017.12 | 67.8 | 73.4 | 62.9 | 32.3 |