“…For the generation of court view, we can take them as text generation problems, such as BLEU-1, BLEU-2, BLEU-N, ROUGE-1, ROUGE-2, ROUGE-L and BERT SCORE [71,72,73,74,34,75,76,77]. The following evaluation metrics to evaluate the performance of LJP text generation are summarized, as shown in Fig.…”