An Approach Based on Multilevel Convolution for Sentence-Level Element Extraction of Legal Text

Table 5

The results on Labor dataset test dataset.

Models	HL (-)	Micro-P (+)	Micro-R (+)	Micro-F1 (+)
CNN	0.02499071	0.8469	0.8275	0.8371
CNN-RNN	0.02551891	0.8497	0.8365	0.8430
SGM	0.01820437	0.8740	0.8451	0.8593
SU4MLC	0.01758433	0.8796	0.8487	0.8639
ML-Reasoner	0.01715832	0.8759	0.8633	0.8695
Our model	0.01688988	0.8812	0.8589	0.8699