📰 Fake News Detection using Transformer-BiGRU

This project implements a Fake News Detection system using a hybrid deep learning model combining a Transformer and a Bidirectional GRU architecture. The model is trained to classify political and news-related statements as true or false using both the LIAR dataset and the ISOT Fake News Dataset.

---

📌 Project Overview

Fake news has become a significant concern in today's media landscape. This project aims to identify misinformation in short political and general news statements using deep learning techniques. It leverages pre-trained word embeddings and advanced sequence models to improve detection accuracy.

---

🧠 Model Architecture

The model architecture includes:

• Embedding Layer: Initialized with pre-trained GloVe (Global Vectors for Word Representation) embeddings.
• Transformer Encoder Block: Captures global contextual relationships.
• Bidirectional GRU Layer: Learns sequential dependencies from both forward and backward directions.
• Dropout Layers: Prevent overfitting.
• Fully Connected Output Layer: Produces classification logits for binary classification.

---

📊 Datasets Used

1. LIAR Dataset

• Source: POLITIFACT
• 12,836 labeled short political statements
• Labels: true, mostly-true, half-true, barely-true, false, pants-fire
• Simplified to binary labels: Real vs. Fake
• Used for training, validation, and testing

2. ISOT Fake News Dataset

• Source: University of Victoria ISOT
• 44,898 news articles:
  • 21,417 real news (from Reuters)
  • 23,481 fake news (from unreliable sources)
• Used for general news-based fake vs real classification

---

⚙️ Preprocessing

• Tokenization using keras.preprocessing.text.Tokenizer
• Padding sequences to a fixed length (e.g., 100 tokens)
• Simplifying multi-class labels (in LIAR) to binary
• GloVe Embedding loading (100-dimensional vectors)

---

🏃‍♂️ Training

• Optimizer: Adam
• Loss: Binary Cross-Entropy
• Metrics: Accuracy
• Epochs: 10–20 (adjustable)
• Validation accuracy typically exceeds 85%+

---

📈 Evaluation

Evaluation is performed using test sets from both datasets. Metrics include:

• Accuracy
• Precision
• Recall
• F1 Score
• Confusion Matrix

---

🛠️ Requirements

Install dependencies with:

pip install tensorflow numpy pandas matplotlib scikit-learn
Make sure the GloVe embeddings (glove.6B.100d.txt) are downloaded and accessible in your project path.