Table 9. Sentence prediction results using different models on both injury and public endangerment cases.

Crime type	Model	Accuracy	$\mathrm{P}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{i}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n}$	$\mathrm{R}\mathrm{e}\mathrm{c}\mathrm{a}\mathrm{l}\mathrm{l}$	$\mathrm{F}1$
Injury	XLNET	70.14%	69.49%	69.96%	66.12%
	BERT	68.82%	68.40%	68.90%	68.58%
	Two-chunking BERT	72.37%	73.22%	72.31%	72.64%
	Three-chunking BERT	71.49%	70.35%	71.54%	69.13%
Public endangerment	XLNET	79.06%	78.30%	79.05%	77.94%
	BERT	73.03%	75.41%	73.03%	73.61%
	Two-chunking BERT	80.16%	79.43%	80.16%	79.34%
	Three-chunking BERT	80.93%	80.53%	80.93%	80.50%

Note:

Bold values indicate the optimal value of the injury and public endangerment evaluation index.