Error Detection on CoNLL-14 (test2)
37.72PrecisionBiLSTM-JOINT
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| BiLSTM-JOINTOptimization=all auxiliary objectives2018.11 | 37.72 | 16.19 | 29.65 | |
| Ann+PAT+MTTraining Data=Manual annotation + Pattern-based AEG + SMT-based AEG2017.07 | 35.28 | 19.42 | 30.13 | |
| BiLSTM-ATTNSupervision=only for sequence labeling2018.11 | 34.13 | 12.76 | 25.22 | |
| Ann+PATTraining Data=Manual annotation + Pattern-based AEG2017.07 | 34.04 | 17.32 | 28.49 | |
| Ann+FY14Training Data=Manual annotation + Felice and Yuan (2014) artificial data2017.07 | 33.25 | 16.67 | 27.72 | |
| Ann+MTTraining Data=Manual annotation + SMT-based AEG2017.07 | 30.24 | 22.96 | 28.39 | |
| Rei (2017)Description=best system from Rei (2017)2018.11 | 27.62 | 21.18 | 25.88 | |
| AnnotationTraining Data=Manual annotation only2017.07 | 25.72 | 20.92 | 24.57 | |
| Rei & Yannakoudakis (2016)Training Data=Provided in original publication2017.07 | 23.6 | 25.1 | 23.9 |