MAOffens

Offensive Language detection in Moroccan Arabic

This work aims to detecte offensive language in Moroccan Arabic used in social media. The first step was the collection and the annotation of a dataset of 23k comments from YouTube written in Moroccan Arabic. Then, an analysis was conducted to explore the data as well as the annotation. Finally, we conducted a series of experiments to evaluate the performance of the dataset in offensive language classification.