This project implements a deep learning model to detect spam questions on Quora using a bidirectional LSTM architecture.

• spam_filter_quora.ipynb: Main notebook containing the model implementation
• requirements.txt: List of Python dependencies
• .gitignore: Git ignore rules
• README.md: Project documentation

1. Create a virtual environment:

 python -m venv .venv
 source .venv/bin/activate  # On Windows: .venv\Scripts\activate

2. Install dependencies:

pip install -r requirements.txt

3. Download GloVe embeddings:

Download glove.6B.300d.txt from Stanford NLP website

• Bidirectional LSTM with GloVe embeddings (300d)
• Multiple LSTM layers with dropout (0.5)
• Dense layers for classification
• Global Average Pooling
• Adam optimizer with learning rate scheduling

This repository includes:

• train.csv: Training dataset containing Quora questions with following columns:

  • qid: Unique question identifier
  • question_text: The actual question text
  • target: Binary label (0 for non-spam, 1 for spam)

• Initial class distribution:

  • Non-spam (0): 1,225,312
  • Spam (1): 80,810

• Balanced using RandomOverSampler to handle class imbalance

Classification Report (with threshold = 0.9):

• Best F1 Score: 0.989
• Final Validation Accuracy: 98.83%
• Final Validation AUC: 0.991

• Maximum words: 30,000
• Maximum sequence length: 100
• Embedding dimension: 300
• LSTM units: 128
• Dense units: 64
• Dropout rate: 0.5
• Learning rate: 0.001 (with reduction on plateau)