• Sakshi Sahu
• Prof Nitya Khare

When it comes to safeguarding information and maintaining public discourse, fake news has emerged as a major concern. This paper proposes a very effective automated approach for detecting false news that relies on a deep sequential model that utilizes “Long Short-Term Memory (LSTM)” networks. A robust preprocessing workflow and a two-layer LSTM architecture with dropout regularization were used on the 44,898 articles that made up the dataset. On the test set, the model outperformed conventional machine learning methods with an astounding accuracy of 98.71% with precision, recall, and F1-score values of 0.99. Also, by a wide margin of 23.71 percentage points, it beat the best-performing Logistic Regression model that had been previously documented, which had grid search optimization and only reached 75% accuracy. Deep sequential models have successfully identified the misleading content's linguistic patterns and contextual relationships, according to the results. For this reason, they have recently been hailed as an improved and scalable alternative to current methods for detecting false news in the real world.

Fake news detection, deep sequential models, LSTM, natural language processing, misinformation classification, deep learning, text classification, recurrent neural networks.

"Deep Sequential Modeling for Robust Fake News Classification", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 12, page no.f340-f347, December-2025, Available :http://www.jetir.org/papers/JETIR2512544.pdf

"Deep Sequential Modeling for Robust Fake News Classification", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 12, page no. ppf340-f347, December-2025, Available at : http://www.jetir.org/papers/JETIR2512544.pdf