HYBRID DEEP LEARNING FRAMEWORK FOR EARLY DETECTION OF FAKE NEWS IN SOCIAL MEDIA PLATFORMS

Abstract

The rapid growth of social media platforms has significantly accelerated the spread of information
across the globe. However, this accessibility has also facilitated the widespread dissemination of
misinformation and fake news, which can negatively influence public opinion, political stability,
and social harmony. Detecting fake news at an early stage is therefore a critical challenge for
modern information systems. Traditional machine learning approaches often struggle to capture
complex linguistic patterns and contextual relationships present in social media content. To
address this issue, this paper proposes a Hybrid Deep Learning Framework for Early Detection of
Fake News in Social Media. The proposed framework integrates semantic feature extraction using
Bidirectional Encoder Representations from Transformers, contextual sequence modeling through
Long Short-Term Memory, and feature refinement using Convolutional Neural Network layers.
This hybrid architecture captures both global contextual semantics and local linguistic patterns
present in news content and social media posts. The system performs data preprocessing, feature
extraction, hybrid model training, and classification to distinguish between fake and genuine news
articles. Experimental evaluation demonstrates that the proposed model achieves improved
detection accuracy, reduced false positives, and faster identification of misinformation compared
with traditional machine learning approaches. The framework is particularly effective for early-
stage detection where limited propagation information is available. The proposed method
provides a scalable and efficient solution for combating misinformation in modern social media
ecosystems.