Emotion Detection on SemEval-2019 Task 3 (test)
0.7382F1 (happy)NELEC
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| NELECData Pre-Processing=Without2019.04 | 0.7382 | 0.8047 | 0.7873 | 0.7765 | — | — | — | — | |
| NELECData Pre-Processing=With2019.04 | 0.7324 | 0.8015 | 0.7878 | 0.7736 | — | — | — | — | |
| LogisticData Pre-Processing=With2019.04 | 0.6782 | 0.768 | 0.712 | 0.7177 | — | — | — | — | |
| LogisticData Pre-Processing=Without2019.04 | 0.6712 | 0.7642 | 0.7151 | 0.7154 | — | — | — | — | |
| DeepData Pre-Processing=Without2019.04 | 0.5863 | 0.5977 | 0.6485 | 0.6123 | — | — | — | — | |
| BaselineData Pre-Processing=With2019.04 | 0.5797 | 0.5973 | 0.6241 | 0.6024 | — | — | — | — | |
| DeepData Pre-Processing=With2019.04 | 0.571 | 0.663 | 0.735 | 0.6651 | — | — | — | — | |
| BaselineData Pre-Processing=Without2019.04 | 0.5461 | 0.6149 | 0.5945 | 0.5861 | — | — | — | — | |
| BERTBackbone=BERT-base2019.03 | — | — | — | — | 0.7151 | 0.7654 | 0.8157 | 0.7631 | |
| HRLCEArchitecture=Hierarchical Recurrent LSTM, Embeddings=Contextualized2019.03 | — | — | — | — | 0.722 | 0.766 | 0.818 | 0.7666 | |
| SA-LSTM (SL)Embeddings=GloVe + ELMo, Encoder=Bidirectional LSTM, Mechanism=Self-attention2019.03 | — | — | — | — | 0.64 | 0.719 | 0.73 | 0.6939 | |
| SA-LSTM-DeepMoji (SLD)Embeddings=GloVe + ELMo + DeepMoji, Encoder=Bidirectional LSTM, Mechanism=Self-attention2019.03 | — | — | — | — | 0.635 | 0.718 | 0.736 | 0.6934 |