iSentenizer-𝝁: Multilingual Sentence Boundary Detection Model

Table 10

Results on the Brown, WSJ, and Tycho Brahe corpus.

Corpus	Candidates	Recall	Precision	𝐹-Score
WSJ corpus	iSentenizer	98.48%	99.18%	98.83%
	Punkt	93.08%	57.84%	71.34%
	MxTerminator	93.08%	60.24%	73.14%
Brown corpus	iSentenizer	99.41%	99.98%	99.70%
	Punkt	96.30%	99.95%	98.09%
	MxTerminator	96.30%	99.98%	98.11%
Tycho Brahe corpus	iSentenizer	99.40%	99.86%	99.63%
	Punkt	79.83%	99.90%	88.74%
	MxTerminator	79.83%	99.98%	88.77%