eISSN 2093-744X
pISSN 1598-2645

Table 6

Performance Comparison in terms of F1 Score and MCC for three different models on four different datasets, before and after augmentation with various sizes and with data duplication technique

Model	Dataset	Before augmentation		Augmentation 10%		Augmentation 20%		Augmentation 30%		Data duplication 20%

		F1	MCC	F1	MCC	F1	MCC	F1	MCC	F1	MCC
BERT	iSarcasm	0.3604	0.2421	0.3450	0.2499	0.3217	0.2567	0.2922	0.2410	0.3406	0.2817
	Ghosh	0.7846	0.6184	0.7760	0.6042	0.7693	0.5963	0.7732	0.6048	0.8042	0.6541
	Ptacek	0.8596	0.7184	0.8602	0.7192	0.8598	0.7190	0.8587	0.7165	0.8643	0.7263
	SemEval-18	0.6442	0.3642	0.6503	0.3825	0.6706	0.4308	0.6257	0.3505	0.6259	0.3578

RoBERTa	iSarcasm	0.3834	0.2842	0.3809	0.2964	0.4044	0.3084	0.3828	0.2939	0.3925	0.2914
	Ghosh	0.7904	0.6299	0.7830	0.6284	0.7758	0.6193	0.7835	0.6294	0.8108	0.6691
	Ptacek	0.8735	0.7454	0.8738	0.7491	0.8727	0.7469	0.8717	0.7442	0.8741	0.7474
	SemEval-18	0.6637	0.4109	0.6666	0.4286	0.6707	0.4362	0.6746	0.4382	0.6627	0.4134

DistilBERT	iSarcasm	0.3083	0.2080	0.2924	0.1890	0.2784	0.1926	0.3399	0.2460	0.2991	0.2224
	Ghosh	0.7831	0.6148	0.7651	0.5854	0.7620	0.5799	0.7571	0.5700	0.7972	0.6403
	Ptacek	0.8542	0.7101	0.8538	0.7094	0.8508	0.7055	0.8569	0.7163	0.8557	0.7118
	SemEval-18	0.6066	0.3180	0.6240	0.3534	0.6522	0.4073	0.6366	0.3822	0.6130	0.3261

International Journal of Fuzzy Logic and Intelligent Systems 2022;22:401~413 https://doi.org/10.5391/IJFIS.2022.22.4.401