Table 3 Main results. The results marked with the asterisks are based on our reproduction. A bold number denotes the highest value in that column. The Bold is the best results.
From: Semantic-enhanced graph neural network for named entity recognition in ancient Chinese books
Groups | Methods | P | R | F |
|---|---|---|---|---|
Without dictionary | CNN-CRF | 56.61 | 57.12 | 56.86 |
BiLSTM-CRF | 56.82 | 57.2 | 57.01 | |
TENER | 57.42 | 59.36 | 58.37 | |
BAC-CRF | 66.12 | 76.42 | 70.89 | |
With dictionary | LR-CNN | 60.62 | 63.65 | 62.10 |
Lattice-LSTM | 60.14 | 62.42 | 61.26 | |
NFLAT | 62.86 | 64.09 | 63.47 | |
Ours | BAC-GNN-CRF | 70.98 | 78.86 | 74.71 |
