Fake News Detection System

A machine learning system that analyzes news article text and classifies it as likely real or fake.

Misinformation spreads quickly online, and manually verifying every article is impractical - there was a need for an automated first-pass classifier.

We trained a text classification model on a labeled news dataset using NLP preprocessing and a Scikit-learn classifier, then wrapped it in a simple Flask web interface for testing article text.

• Text preprocessing and cleaning pipeline (NLTK)
• TF-IDF feature extraction
• Trained classification model (Logistic Regression / Naive Bayes)
• Simple web interface to paste and test article text
• Displays prediction with a confidence score

Raw article text is cleaned and vectorized using TF-IDF, then passed to a trained Scikit-learn classifier. The model is served through a lightweight Flask backend with a simple HTML form for input and results.

• Cleaning noisy, inconsistent text data from the training dataset
• Balancing the dataset to avoid biased predictions
• Explaining model limitations clearly, since no classifier is 100% accurate