The rapid spread of fake news and misinformation across digital platforms has emerged as a serious
challenge to public trust and informed decision-making. This project focuses on addressing the problem
of automated fake news detection by designing an efficient text classification system. The proposed
approach examines a given news article and determines whether it is classified as fake or true based
solely on its textual information. The foundation of the system is a pre-trained transformer-based
language model, DistilBERT, which is capable of producing rich contextual and semantic representations
of text. Before processing the input through the model, the text undergoes essential Natural Language
Processing (NLP) preprocessing to ensure standardized and noise-free input. DistilBERT, a compact and
computationally optimized variant of BERT, is fine-tuned for binary classification, achieving a balance
between high predictive accuracy and fast inference. The model effectively identifies subtle linguistic
patterns associated with misinformation. The final application offers a scalable and reliable solution
for fake news detection, contributing to efforts aimed at reducing the impact of digital misinformation.
