Detecting offensive speech in Arabic in social media.
Through a wide variety of mobile applications, we’ve developed a unique visual system and strategy that can be applied across the spectrum of available applications.
the main goal of this project is to classify offensive content written in Arabic dialect (Egyptian dialect, Levantine dialect, Gulf dialect, and Moroccan dialect) on Facebook and visualize the results.
Offensive content is content that reasonably causes another to experience extreme anger, insult, or disrespect. Examples of offensive content
- Content that is threatening, harassing, or incites violence.
Hate speech often emerges from an “us vs. them” conceptual framework, in which individuals differentiate the group they believe they belong to, or the “in-group,” from the “out-group.” Hate speech toward the out-groups is segmented into three major categories in this analysis.
Dehumanization involves belittling groups and equating them to culturally despised subhuman entities, such as pigs, rats, monkeys, or even germs or dirt/filth. A widely known recent manifestation of this phenomena involved calling the Tutsi minority in Rwanda cockroaches in the lead up and during the 1994 genocide.
Demonization, on the other hand, involves portraying a group as superhuman, like a monster, robot, or even fatal diseases like cancer that are a mortal threat to the in-group.
- Racism
- Sexism (Misogyny,Misandry)
- religion off (againt religion)