Hate Speech Detection towards the Mexican Spanish Speaking LGBT+ Population (HOMO-MEX)

The HOMO-MEX task (webpage) presented at IberLEF 2023 focused on detecting LGBTQ+ phobic content in Spanish tweets.

The following code can generate an instance of the system used in the competition.

>>> from EvoMSA.competitions import Comp2023
>>> D = # Training set
>>> tailored = 'IberLEF2023_HOMO-MEX'
>>> comp2023 = Comp2023(lang='es', tailored=tailored)
>>> ins = comp2023.stack_3_bow_tailored_all_keywords(D)
Performance in Cross-validation

Configuration

Performance

p-value

Comp2023.stack_3_bow_tailored_all_keywords

0.7914

1.0000

Comp2023.stack_2_bow_tailored_all_keywords

0.7912

0.4460

Comp2023.stack_2_bow_tailored_keywords

0.7908

0.3420

Comp2023.stack_2_bow_keywords

0.7904

0.2980

Comp2023.stack_2_bow_all_keywords

0.7903

0.2700

Comp2023.stack_3_bows_tailored_keywords

0.7901

0.0740

Comp2023.stack_bow_keywords_emojis_voc_selection

0.7885

0.1300

Comp2023.stack_bows

0.7880

0.1460

Comp2023.stack_bow_keywords_emojis

0.7871

0.0660

Comp2023.stack_3_bows

0.7861

0.0160

Comp2023.bow_voc_selection

0.7689

0.0000

Comp2023.bow

0.7669

0.0000

Comp2023.bow_training_set

0.7553

0.0000