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

Table 3

ROGUE2 comparisons between different methods.

Method\value	P	R	F1
LEAD	14.1	12.4	13.2
NEURALSUM	14.3	13.3	13.8
TextRank	13.6	16.9	15.1
BERTSUM + FC	12.7	16.8	14.5
BERTSUM + Transformer	13.2	18.1	15.3
BERTSUM + LSTM	12.3	18.5	14.8
BERTSUM + BiLSTM + attention	13.1	17.5	15.0
BERTSUM + CNN	14.2	15.6	14.9
BERTSUM + BiLSTM + CNN	12.4	21.0	15.6
Proposed method	15.2	17.5	16.3

Notes. The value of the proposed method is the average value after 10-fold cross-validation test.