Sentiment Classification on SemEval Twitter Sentiment Analysis 2013 (test)
72.8Average F1Previous best historical scores
Evaluation results
| Method | Links | |
|---|---|---|
| Previous best historical scores2017.04 | 72.8 | |
| Ensemble modelcomponents=10 CNNs and 10 LSTMs2017.04 | 72.5 | |
| CNNEmbedding=fasttext, Convolution size=[3,4,5]2017.04 | 72 | |
| LSTMEmbedding=word2vec2017.04 | 72 | |
| LSTMEmbedding=word2vec, Fully connected layer=false2017.04 | 71.9 | |
| CNNEmbedding=word2vec, Convolution size=[3,4,5]2017.04 | 71.5 | |
| CNNEmbedding=word2vec, Convolution size=[3,4,5], Fully connected layer=false2017.04 | 71.5 | |
| CNNEmbedding=word2vec, Convolution size=[1,2,3]2017.04 | 71.2 | |
| LSTMEmbedding=fasttext2017.04 | 71.2 | |
| CNNEmbedding=word2vec, Convolution size=[5,6,7]2017.04 | 71 | |
| LSTMEmbedding=glove2017.04 | 71 | |
| CNNEmbedding=glove, Convolution size=[3,4,5]2017.04 | 70.9 | |
| CNNEmbedding=word2vec, Convolution size=[3,4,5], Distant training=false2017.04 | 69.8 | |
| LSTMEmbedding=word2vec, Distant training=false2017.04 | 69.8 | |
| LSTMEmbedding=word2vec, Class weights=false2017.04 | 68.9 | |
| CNNEmbedding=word2vec, Convolution size=[3,4,5], Class weights=false2017.04 | 68.2 | |
| Logistic regressionbaseline=1-3 grams2017.04 | 62.7 |