News Classification Using Deep Learning Techniques and NLP

Harsh Kansal, Chandan Tyagi, Himani Chaudhary, Chandan Tyagi, Prabha Tarar, Amiksha Dixit

Abstract: This paper proposes a news classification system using Natural Language Processing and deep learning. The news categorization system classifies articles into one of the five classes, namely, Politics, Defense, Sports, Entertainment and Technology. Any article that does not belong to the first three classes is classified as ?Out-of-Scope.? We employ a fine-tuned Bidirectional Encoder Representations from Transformers model enhanced with Bidirectional Long Short-Term Memory layers to achieve multi-class text classification robustly. The system is tested on a handpicked dataset collected from several publicly available news corpora and has a macro-averaged F1-score of 0.945. The transformer-based models we trained perform significantly better than traditional machine learning (ML) approaches. This finding corroborates results in recently published literature. The proposed rejection mechanism for off-topic articles adds practical value for deployment.

Keywords: News classification, deep learning, NLP, BERT, BiLSTM, text categorization, multi-class classification, out-of-scope detection

How to Cite?: Harsh Kansal, Chandan Tyagi, Himani Chaudhary, Chandan Tyagi, Prabha Tarar, Amiksha Dixit, "News Classification Using Deep Learning Techniques and NLP", Volume 15 Issue 4, April 2026, International Journal of Science and Research (IJSR), Pages: 1553-1556, https://www.ijsr.net/getabstract.php?paperid=SR26422170242, DOI: https://dx.doi.org/10.21275/SR26422170242

Download Citation: APA | MLA | BibTeX | EndNote | RefMan