Table 5 The benchmark comparison for Kaggle dataset. The bold font indicates the highest BA within each respective dimension in the experiments.

Majority Baseline

0.8617

0.6040

0.7689

0.5412

0.6939

⁴¹ EGB

-

0.8606

-

0.6570

-

0.7817

-

0.7178

-

-

-

⁴¹ RNN

-

0.6200

-

0.6370

-

0.6760

-

0.7780

-

-

-

³ LIWC-MLP

-

0.8630

-

0.6190

-

0.7760

-

0.7200

-

0.7445

-

-

³ BERT-MLP

-

0.8630

-

0.6720

-

0.7880

-

0.7610

-

-

-

⁴ Oversampling

-

-

-

-

-

-

-

-

-

-

-

0.1440

⁴ SMOTE

-

-

-

-

-

-

-

-

-

-

-

0.2870

Baseline \(\mathscr {L}_{\text {BCE}}\)

0.6599

0.8720

0.6979

0.7159

0.7074

0.8236

0.8049

0.8069

0.7175

0.8046

0.7419

0.3059

Proposed \(\mathscr {L}_{\text {FBCE-W}}\)

0.7757

0.7942

0.7148

0.7239

0.7617

0.7850

0.8131

0.8161

0.7663

0.7798

0.7422

0.3561

Proposed \(\mathscr {L}_{\text {FBCE-T}}\)

0.7979

0.8386

0.7366

0.7245

0.7885

0.8236

0.8372

0.8375

0.7901

0.8061

0.7749

0.3901

Proposed \(\mathscr {L}_{\text {FBCE-M}}\)

0.7970

0.8461

0.7306

0.7349

0.7873

0.8311

0.8352

0.8357

0.7875

0.8120

0.7730

0.3972