Detecting and Analyzing Harmful Behavior on Social Media

With the massive increase in the use of social media platforms such as Facebook, a new era of information warfare on people or states has begun (Facebook, Twitter, Instagram, Snapchat, Whatsapp, etc.). False information (e.g., rumors, fake news, hoaxes, clickbait, and other harmful behavior) can change the public's opinion on certain topics, when celebrity actions, bullying, or hate speech on social media can harm children's and teenagers' behavior. A machine learning model requires features that classified accurately to a certain classification to detect dangerous conduct. In this paper, we use a natural language processing-based convolutional neural network model to classify multiple Arabic datasets from different social media websites. These crucial procedures entail selecting the proper dataset and getting the dataset ready for public consumption. It seeks to provide a focus and a place to begin developing an effective algorithm that performed posting-related natural language processing to safeguard Arab users from negative behavior on various social media platforms. The results are encouraging and demonstrate a high level of bad emotional behavior recognition.