Indian Fake News Detection System using Deep Learning Model and Adaptive Learning

The proliferation of fake news poses significant challenges to societal well-being and the integrity of information. This paper presents a scalable and robust modular system for fake news detection, leveraging advanced deep learning models integrated with Named Entity Recognition (NER), claim extraction, and adaptive learning. By utilizing a fine-tuned DistilBERT model, the system achieves a state-of-the-art accuracy of 98%, surpassing existing methods. Key contributions include structured insights through NER and claim extraction for improved transparency and interpretability, alongside an adaptive learning mechanism that ensures scalability across domains and adaptability to evolving misinformation patterns. The proposed system’s efficiency and domain robustness make it a practical solution for combating fake news, particularly in diverse and data-intensive contexts such as India. Future work includes exploring hybrid generative-discriminative model synergies to further enhance system performance.