Automatic Generation of the Draft Procuratorial Suggestions Based on an Extractive Summarization Method: BERTSLCA

Table 7

Performance on CAIL 2020 dataset.

Method\value	ROGUE1-F1	ROGUE2-F1	ROGUEL-F1
LEAD	18.5	12.4	18.4
NEURALSUM	25.4	13.3	19.8
TextRank	22.5	16.9	17.6
BERTSUM + FC	25.7	16.8	22.0
BERTSUM + Transformer	26.5	18.1	22.7
BERTSUM + LSTM	26.1	18.5	23.6
BERTSUM + BiLSTM + attention	28.2	17.5	24.1
BERTSUM + CNN	28.9	15.6	23.2
BERTSUM + BiLSTM + CNN	29.4	21.0	25.8
Proposed method	29.8	20.5	29.1