. 2021 Oct 19;7(10):e08216. doi: 10.1016/j.heliyon.2021.e08216

Table 15.

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

Input	Model	Micro-averaged					Macro-averaged
Input	Model	P	R	$F_{1}$ (Mean)	$F_{1}$ (SD)	$F_{1}$ (Max)	P	R	$F_{1}$ (Mean)	$F_{1}$ (SD)	$F_{1}$ (Max)
Original	CNN	0.7793	0.7793	0.7793	0.0119	0.7989	0.7790	0.7759	0.7756	0.0121	0.7962
	BiLSTM	0.7229	0.7229	0.7229	0.0245	0.7542	0.7323	0.7205	0.7183	0.0251	0.7517
	Hybrid	0.7268	0.7268	0.7268	0.0210	0.7542	0.7397	0.7212	0.7231	0.0203	0.7512
+all POS Tags	CNN	0.7888	0.7888	0.7888	0.0195	0.8156	0.7917	0.7865	0.7862	0.0203	0.8142
	BiLSTM	0.7523	0.7523	0.7523	0.0357	0.7598	0.7583	0.7475	0.7476	0.0339	0.7592
	Hybrid	0.7467	0.7467	0.7467	0.0312	0.7542	0.7520	0.7528	0.7463	0.0300	0.7522
+focusing Tags	CNN	0.7955	0.7955	0.7955	0.0188	0.8324	0.7976	0.7929	0.7932	0.0189	0.8309
	BiLSTM	0.7162	0.7162	0.7162	0.0111	0.7263	0.7212	0.7133	0.7120	0.0119	0.7317
	Hybrid	0.7503	0.7503	0.7503	0.0227	0.7877	0.7540	0.7467	0.7441	0.0208	0.7758
+focusing Words	CNN	0.7313	0.7313	0.7313	0.0107	0.7347	0.7266	0.7271	0.7542	0.0102	0.7471
	BiLSTM	0.6777	0.6777	0.6777	0.0318	0.7207	0.6834	0.6748	0.6740	0.0320	0.7161
	Hybrid	0.7000	0.7000	0.7000	0.0225	0.7207	0.7267	0.7155	0.7191	0.0225	0.7191