Automatic Conspiracy Theory Identification (ACTI)

The ACTI task (webpage) presented at EVALITA 2023 focused on the automatic identification of conspiracy theories.

>>> from EvoMSA.competitions import Comp2023
>>> D = # Training set
>>> tailored = 'evalita2023_ACTI'
>>> comp2023 = Comp2023(lang='it', tailored=tailored)
>>> ins = comp2023.stack_3_bows_tailored_keywords(D)

Performance in Cross-validation

Configuration	Performance	p-value
Comp2023.stack_3_bows_tailored_keywords	0.7914	1.0000
Comp2023.stack_2_bow_tailored_keywords	0.7872	0.1180
Comp2023.stack_3_bows	0.7855	0.1700
Comp2023.stack_2_bow_keywords	0.7855	0.0740
Comp2023.stack_bow_keywords_emojis_voc_selection	0.7804	0.0240
Comp2023.stack_bows	0.7785	0.0260
Comp2023.bow_voc_selection	0.7780	0.0200
Comp2023.bow_training_set	0.7765	0.0040
Comp2023.bow	0.7758	0.0040
Comp2023.stack_bow_keywords_emojis	0.7755	0.0000