Logo Lanfrica

xprogramer/OBAM

Domaine:

natural language processing

Type de record:

dataset
Créateur:
xpr
Hôte:
Corpus of offensive Bambara language # OBAM Corpus of offensive Bambara language ## OBAM v1.0 The first version of the corpus contains 4,072 texts written in code-switched Bambara-French, where the texts are labeled as normal (0), offensive (1) and abusive (2). The annotation has been done by three native Bambara speakers from Mali, where we have applied the majority voting to attribute the final label for each text. | Category | #texts | | --------- | -------| | normal | 2,835 | | offensive | 695 | | abusive | 541 | | Total | 4,072 | Please, cite the following article for research purpose: https//