Table 1. Selected examples of NLP tasks that can be covered by the method.

Task	T	F	T′	Y
Sentiment classification (model 1)	“Great music!” Nt = 2	A feature set for each token, Nf = 2	No = 0	‘YES’ Ny = 1
Sentiment classification (model 2)	“Great music!” Nt = 2	A feature set for the whole sentence, Nf = 1	No = 0	‘YES’ Ny = 1
Word sense disambiguation	“We bought gas for the car.” Nt = 6	A feature set for each token, Nf = 6	No = 0	Sense = automobile Ny = 1
Named Entity Recognition (in Turkish)	“Henüz Ali Sami Yen Stadyumu taşınmamıştı.”, Nt = 6	A feature set for each token, Nf = 6	No = 0	“O B-LOC I-LOC I-LOC E-LOC O” Ny = Nt = 6
Machine translation (from Turkish to English)	“Henüz Ali Sami Yen Stadyumu taşınmamıştı.”, Nt = 6	A feature set for each token, Nf = 6	“Ali Sami Yen Stadium was not relocated yet.” No = 8	Ny = 0
Morphological disambiguation (in Turkish)	“Henüz Ali Sami Yen Stadyumu taşınmamıştı.”, Nt = 6	A feature set for each token, Nf = 6	“Henüz Ali Sami Yen Stadyum+u taşın+ma+mış+tı” No = 6	“Henüz Ali Sami Yen Stadyum+Acc taşın+Neg+PastPart+Past”, Ny = 6