Table 4.
Comparison of method performance on the China Conference on Knowledge Graph and Semantic Computing 2018 dataset.
| Method | P value (%) | R value (%) | F1 score (%) |
| BiLSTMa-CRFb [17] | 65.68 | 69.04 | 67.32 |
| SMc-LSTM-CRF [18] | 80.54 | 79.61 | 80.08 |
| Multi-level representation learning for entity recognition (multi-level ER) | 83.90 | 82.47 | 83.18 |
aBiLSTM: bidirectional long short-term memory.
bCRF: conditional random field.
cSM: self-matching attention mechanism.