Table 9.

Comparing F₁-score of traditional models (Thai sentences dataset) with focusing POS tags (N+Adj+Det: NAD).

		Rang of N-gram
		Micro $F_1$-score					Macro $F_1$-score
Dataset/Model		Unigram	Uni+ Bigram	Uni+ Trigram	Bigram	Trigram	Unigram	Uni+ Bigram	Uni+ Trigram	Bigram	Trigram
Original	MNB	0.6425	0.6704	0.6592	0.6648	0.5419	0.6375	0.6726	0.6585	0.6652	0.5344
	SVM	0.7374	0.7318	0.7318	0.6592	0.5363	0.7394	0.7361	0.7331	0.6583	0.5289
	LR	0.7095	0.7039	0.6927	0.6760	0.5307	0.7062	0.7007	0.6898	0.6758	0.5229
+all POS Tags	MNB	0.6760	0.6872	0.6480	0.6927	0.6089	0.6673	0.6827	0.6408	0.6855	0.5959
	SVM	0.7430	0.7374	0.7263	0.7095	0.6201	0.7430	0.7395	0.7253	0.7075	0.6088
	LR	0.7207	0.7318	0.6816	0.6760	0.5866	0.7188	0.7319	0.6741	0.6703	0.5681
+focusing Tags	MNB	0.6872	0.6872	0.7039	0.6927	0.6034	0.6809	0.6809	0.6954	0.6885	0.5856
	SVM	0.7374	0.7374	0.7486	0.7263	0.6201	0.7382	0.7382	0.7503	0.7257	0.6055
	LR	0.7598	0.7598	0.7095	0.6704	0.5866	0.7599	0.7599	0.7088	0.6695	0.5684
+focusing Words	MNB	0.6318	0.6536	0.6536	0.6425	0.5307	0.6320	0.6541	0.6534	0.6410	0.5247
	SVM	0.6983	0.6760	0.6704	0.6369	0.5307	0.7001	0.6752	0.6688	0.6358	0.5247
	LR	0.6872	0.6760	0.6536	0.6257	0.5251	0.6835	0.6746	0.6505	0.6208	0.5177