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

Table 1

Statistics of dataset.

Dataset	Total labels	Sample of train set	Sample of validation set	Sample of test set	Words/sample	Label/sample
RCV1V2	103	802414	1000	1000	123.94	3.24
Divorce	20	12727	3927	3932	104.28	1.75
Loan	20	6384	1993	1975	169.55	1.93
Labor	20	6045	1855	1837	131.28	1.51