Unveiling Fake News with an Attention-Based CNN Model Using Hybrid Features

Sophisticated detection mechanisms are required to protect the integrity of information in the digital age, as the widespread dissemination of false news on social media platforms has become a critical challenge. To tackle the practical issue of spreading false information, there is an urgent need for reliable systems that can effectively identify and counteract the propagation of fake news. This work presents a novel method for identifying fake news by combining sophisticated encoding techniques and utilizing deep learning architectures. Our approach integrates TF-IDF, CountVectorizer (CV), and Word2Vec (W2V) techniques for encoding. These techniques are coupled sequentially to capture the subtle details in fake news content successfully. Through the utilization of attention-based Convolutional Neural Network (CNN) models, we improve the comprehension and categorization of intricate textual material. Our methodology prioritizes scalability to efficiently analyze big datasets while retaining high detection accuracy in many situations and languages. Our approach has been thoroughly tested and proven to be strong and applicable in a wide range of false news situations. We have attained a state-of-the-art level of precision, with an accuracy of 99.11 %, by employing an attention-based Convolutional Neural Network (CNN) coupled CountVectorizer and Word2Vec embeddings,