FANDOM


OntoNotes 5.0 (CoNLL 2012) Edit

Using development and test set specified by CoNLL-2012 Shared Task (coreference resolution).

"Comp." = "complete argument structure"

Model P R F1 Comp. Notes Performance Ref. System Ref.
ASSERT 78.5 76.6 77.5 55.8 Revised numbers first: Pradhan et al. (2013)[1] revised: Täckström et al. (2015)[2] Pradhan et al. (2005)[3]
Täckström et al. (2015) 80.6 78.2 79.4 61.8 Dynamic programming, globally-normalized Täckström et al. (2015)[2] Täckström et al. (2015)[2]
Zhou and Xu (2015)[4] 81.27 Recurrent NN, no syntax Zhou and Xu (2015)[4] Zhou and Xu (2015)[4]

CoNLL-2005 Edit

Model P R F1 Comp. Notes Performance ref. Model ref.
Zhou and Xu (2015)[4] 81.07 Recurrent NN, no syntax Zhou and Xu (2015)[4] Zhou and Xu (2015)[4]

References Edit

  1. Sameer Pradhan, Alessandro Moschitti, Nianwen Xue, Tou Hwee Ng, Anders Björkelund, Olga Uryupina, Yuchen Zhang, and Zhi Zhong. 2013. Towards robust linguistic analysis using OntoNotes. In Proceedings of CoNLL.
  2. 2.0 2.1 2.2 Täckström, Oscar, Kuzman Ganchev, and Dipanjan Das. "Efficient inference and structured learning for semantic role labeling." Transactions of the Association for Computational Linguistics 3 (2015): 29-41.
  3. Sameer Pradhan, Kadri Hacioglu, Valerie Krugler, Wayne Ward, James Martin, and Dan Jurafsky. 2005. Support vector learning for semantic argument classification. Ma- chine Learning, 60(1):11–39. 
  4. 4.0 4.1 4.2 4.3 4.4 4.5 Zhou, Jie, and Wei Xu. "End-to-end learning of semantic role labeling using recurrent neural networks." Proceedings of the Annual Meeting of the Association for Computational Linguistics. 2015.