Table 3.

F1-score of the five different machine learning algorithms (LR, LSTM, BERT, PM-LSTM, and PM-BERT) for the ten different categories (business, computers, games, health, home, recreation, science, shopping, society, and sports).

Model	Category	Cul	Eco	Pol	Ling	Geo	Category	Cul	Eco	Pol	Ling	Geo
LR	Business	0.40	0.48	0.71	0.73	0.61	Recreation	0.37	0.30	0.59	0.73	0.57
	Computers	0.42	0.35	0.58	0.63	0.56	Science	0.42	0.42	0.62	0.71	0.65
	Games	0.52	0.35	0.59	0.59	0.60	Shopping	0.36	0.27	0.49	0.61	0.52
	Health	0.36	0.40	0.58	0.67	0.64	Society	0.40	0.45	0.62	0.68	0.62
	Home	0.39	0.57	0.49	0.68	0.59	Sports	0.44	0.28	0.59	0.62	0.57
LSTM	Business	0.19	0.28	0.49	0.55	0.47	Recreation	0.20	0.39	0.48	0.41	0.47
	Computers	0.20	0.08	0.49	0.52	0.46	Science	0.20	0.20	0.48	0.43	0.43
	Games	0.14	0.29	0.74	0.70	0.48	Shopping	0.23	0.44	0.49	0.44	0.49
	Health	0.18	0.17	0.49	0.59	0.53	Society	0.20	0.27	0.49	0.46	0.48
	Home	0.19	0.21	0.49	0.59	0.63	Sports	0.26	0.43	0.48	0.43	0.49
BERT	Business	0.42	0.54	0.62	0.74	0.50	Recreation	0.32	0.29	0.74	0.81	0.59
	Computers	0.38	0.30	0.47	0.75	0.66	Science	0.39	0.47	0.68	0.78	0.60
	Games	0.40	0.65	0.77	0.85	0.68	Shopping	0.34	0.44	0.49	0.63	0.50
	Health	0.38	0.54	0.66	0.79	0.67	Society	0.39	0.51	0.49	0.73	0.48
	Home	0.32	0.60	0.54	0.81	0.67	Sports	0.48	0.50	0.48	0.76	0.66
PM-LSTM	Business	0.19	0.28	0.49	0.43	0.48	Recreation	0.21	0.47	0.48	0.44	0.46
	Computers	0.21	0.06	0.49	0.46	0.47	Science	0.19	0.21	0.48	0.74	0.43
	Games	0.18	0.22	0.48	0.40	0.48	Shopping	0.18	0.50	0.49	0.50	0.48
	Health	0.20	0.27	0.48	0.43	0.46	Society	0.19	0.25	0.49	0.64	0.48
	Home	0.20	0.27	0.49	0.45	0.48	Sports	0.27	0.22	0.48	0.44	0.48
PM-BERT	Business	0.48	0.66	0.97	0.80	0.96	Recreation	0.66	0.35	0.74	0.48	0.65
	Computers	0.46	0.28	0.54	0.97	0.70	Science	0.47	0.53	0.97	0.78	0.66
	Games	0.33	0.72	0.48	0.47	0.72	Shopping	0.45	0.50	0.97	0.67	0.56
	Health	0.46	0.60	0.72	0.82	0.97	Society	0.52	0.69	0.55	0.73	0.97
	Home	0.41	0.66	0.54	0.83	0.73	Sports	0.49	0.50	0.54	0.79	0.63

The bold values indicate the highest F1-scores that have been achieved by a method for a category.