A Chinese Named Entity Recognition Model of Maintenance Records for Power Primary Equipment Based on Progressive Multitype Feature Fusion

Table 3

Performance comparison of different methods on MSRA dataset.

Methods	Metrics
	P (%)	R (%)	F1 (%)
Dong et al. [30]	91.28	90.62	90.95
Lattice-LSTM-CRF [31]	93.57	92.79	93.18
Cao et al. [32]	91.73	89.58	90.64
WC-LSTM [33]	94.58	92.91	93.74
CNN-BiLSTM-CRF [34]	91.63	90.56	91.09
Proposed method	97.85	95.74	96.63

Bold values show the best performance.