Detecting Illegal Drug Ads and Suppliers across Social Media

Teenagers and young people today spend a lot of time on social media. The youth who use social media have a higher chance of using alcohol, tobacco, and illegal drugs due to various sources available online. Teens' risk of substance use and addiction may be reduced by limiting their exposure to social media posts on illegal drug use and promotion. These platforms have become a source for buying and selling illicit drugs online. Colloquial language used to caption the images associated with drugs is one of the challenging issues for filtering out such posts on social media. In this paper, we offer a technique for automatically identifying social media posts related to illicit drug promotion. With the help of this technique, social media may automatically filter out anything that is associated with illicit drugs. The proposed model uses state-of-the-art social media analytics, which combines text and image processing based on neural networks, to find posts related to illicit drug promotion on social media. Bert tokenizer is used to extract textual features and VGG16 is used for feature extraction from images. Evaluation results show that textual features produced from word embedding and image features derived from VGG16 neural network, when combined together shows better accuracy in classifying posts related to illicit drug promotion compared to other statistical models