DETECTING DEEP FAKE TWEETS WITH FAST TEXT EMBEDDINGS AND CNN-DRIVEN RANDOM FORESTS

ABSTARCTThis research presents an advanced framework for detecting fake tweets using a dedicated Twitter fake tweet dataset.The goal is to improve detection accuracy by combining natural language processing (NLP) and machine learning techniques.Traditional manual detection methods face several limitations, including time-consuming analysis, susceptibility to subjective bias, and poor scalability in the face of the rapid growth of social media data.These methods often fail to identify subtle linguistic and contextual cues associated with misinformation.The proposed approach addresses these challenges through a multi-step pipeline.It begins with NLP preprocessing to clean and standardize the tweets, followed by FastText embeddings to transform textual data into numerical vectors for detailed analysis.The dataset is then split into training and testing sets to ensure robust evaluation.A deep learning convolutional neural network (DLCNN) is used for feature extraction, capturing complex patterns within the data.These features are then classified using a Random Forest classifier to determine whether tweets are real or fake.The model's performance is rigorously evaluated using appropriate metrics, demonstrating its effectiveness and reliability in real-world scenarios.