Scam Calls Detection Using Machine Learning Approaches

In recent years, scam calls have become increasingly prevalent resulting in financial loss, identity theft, and other fraudulent activities. This research proposes a machine learning-based approach for scam call classification and detection using natural language processing (NLP) along with deep learning techniques. The model uses the dataset of scam and non-scam calls to train and understand the context of the caller and determine if the conversation is a scam or not. NLP techniques are leveraged, such as preprocessing text, converting audio samples to texts with Google API, and word embeddings, to build an accurate and reliable classifier. The highest results obtained is Long Short-Term Memory (LSTM) algorithm with an accuracy of 85.61% in detecting scam calls.