Combating Online Fraud: AI-Driven Fake Profile Detection

Since the vast platforms of social media having a wide range of users include a lot of scammers who claim to be genuine but are fake, so to eradicate these spammers it is essential to detect them first. Hence, a deep learning model has been built that detects these fake users. The model gathers a wide range of data from various social media platforms which include followers count, friends count, languages, location services and other valid parameters based on the platform where the user's profile is registered. All these parameters are typically analysed by different models and trained vigorously to find anomalies that account to fraudulent behaviour by the user. The classification of a user as fake or genuine is done using four different models namely Convolutional Neural Networks (CNN), Artificial Neural Networks (ANN) and models of Recurrent Neural Networks like Long Short-Term Memory (LSTM) and Convolutional Long Short-Term Memory (ConvLSTM). Out of these ConvLSTM algorithm(model) used to detect real or fake profiles. In comparison, ConvLSTM topped the competition with a remarkable accuracy of 99.4% followed by LSTM and ANN around 99.3% and 99.1 % respectively whereas CNN had the least accuracy of 98% in detecting the imposters.