Research Article
A Semieager Classifier for Open Relation Extraction
Table 2
Performance comparison on SemEval-2010 Task 8.
| Model | Additional Information | F1 | Epochs (Training) |
| BRNN | Word embeddings | 82.8 | 20 |
| CR-CNN | + | 84.1 | 10 |
| SDP-LSTM | +POS+GR+WordNet embeddings | 83.7 | 110 |
| BLSTM | +PF+POS+NER | 84.3 | 50 |
| Att-BLSTM | Word embeddings + | 84.0 | 100 |
| EAtt-BiGRU | Word embeddings + PF | 84.7 | 10 |
| SemiE | + vector_sum | 84.1 | 1 | + max_polling | 84.6 | 1 | +attention_mode | 85.1 | 1 |
|
|
1: position feature. 2: position indicator, to be inserted into input sentence to specify the boundaries of target entity pair.
|