Violent Speech Detection in Educational Environments

Nowadays, social networks allow people to interact by exchanging messages, publishing public or private photos or videos. But sometimes they become a space of toxic language to criticise, insult, hate, and attack. In this context, researchers promote a strong intention to study, analyse and detect hate speech. By automating its detection, the spread of anxiety and the rise of hateful content can be limited, especially among children in the online schools. However, with the absence of online database of vulgar English speech by Students in schools, detecting violent speech becomes a difficult task. In this paper, we propose a new dataset-based framework for the detection of students violent speech using natural language processing and learning techniques. This focuses on a proposed ≪Students’s Violent Speech (SVS) dataset ≫ with 7056 tagged tweets. The dataset is collected and pre-processed to be analyzed to show the performance and accuracy of the proposed model.