Fake_news_detection

Inspiration

The project is inspired by the increasing spread of misinformation online and the need for automated systems that can help users quickly assess the credibility of news content.

What it does

Identifies the actual context of news and distinguishes between the real and fake news and then gives prediction

How we built it

Advanced — Fine-tuned BERT (bert-base-uncased) A transformer model that understands contextual meaning.

Preprocessing (Light):

• URL and HTML removal
• Whitespace normalization
• No stemming or stopword removal — BERT handles this internally via WordPiece tokenization

Why BERT?

• Bidirectional attention captures full sentence context
• WordPiece tokenizer handles out-of-vocabulary words
• Pre-trained on massive corpora, fine-tuned on our task

Built With

• colab
• github
• huggingface
• kaggle
• matplotlib
• numpy
• pandas
• python
• re
• scikit-learn
• tokenizer
• torch
• transformer