🔍 AI Content Detector Pro

A sophisticated web application that uses advanced algorithms to detect whether text content was written by a human or generated by an AI model. The system employs multiple detection techniques including stylometric analysis, perplexity scoring, and machine learning classification for high-accuracy results.

✨ Features

🧠 Advanced Detection Methods

• Combined Analysis: Integrates all methods for highest accuracy
• Stylometric Analysis: Analyzes writing style patterns and linguistic features
• Perplexity Analysis: Measures text complexity and predictability
• ML Classification: Machine learning-based detection using trained models

📊 Comprehensive Analysis

• Real-time Processing: Instant analysis with detailed results
• Confidence Scoring: Probability-based predictions with confidence levels
• Feature Breakdown: Detailed analysis of text characteristics
• Visual Analytics: Interactive charts and graphs for result visualization

📁 File Support

• Text Files (.txt): Direct text analysis
• Word Documents (.docx): Extract and analyze content
• PDF Files (.pdf): Extract text from PDF documents
• Direct Input: Paste text directly for analysis

🎨 Modern Interface

• Responsive Design: Works on desktop and mobile devices
• Interactive Charts: Plotly-powered visualizations
• Real-time Updates: Live analysis with progress indicators
• Professional UI: Clean, modern interface with intuitive navigation

🚀 Quick Start

Installation

1. Clone the repository:

git clone https://github.com/yourusername/ai-content-detector-pro.git
cd ai-content-detector-pro

2. Create a virtual environment (recommended):

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

3. Install dependencies:

pip install -r requirements.txt

Usage

1. Start the application:

streamlit run app.py

Docker (Share with Friends)

1. Build image

   docker build -t ai-content-detector .

2. Run container

   docker run --rm -p 8501:8501 ai-content-detector

3. Open app

   http://localhost:8501
   

The container already includes all Python/NLTK dependencies, so your friends only need Docker installed.

2. Open your browser and navigate to the provided URL (typically http://localhost:8501)

3. Choose your analysis method:

   • Combined Analysis (Recommended): Uses all methods for best accuracy
   • Stylometric Analysis: Focus on writing style patterns
   • Perplexity Analysis: Analyze text complexity
   • ML Classification: Machine learning-based detection

4. Input your content:

   • Upload a document (.txt, .docx, .pdf)
   • Or paste text directly into the input area

5. Analyze and view results:

   • Get probability scores for human vs AI origin
   • View detailed feature breakdowns
   • Explore interactive visualizations

🔬 Detection Methods Explained

1. Stylometric Analysis 📈

Analyzes writing style characteristics including:

• Vocabulary Richness: Measures diversity of word usage
• Sentence Length Distribution: Analyzes sentence structure patterns
• Word Frequency Analysis: Identifies repetitive patterns
• Punctuation Usage: Examines punctuation patterns
• Capitalization Patterns: Analyzes capitalization frequency
• Word Length Variance: Measures variation in word lengths

2. Perplexity Analysis 🧮

Measures how "surprised" a language model is by the text:

• Higher Perplexity: Suggests human-written content (more unpredictable)
• Lower Perplexity: Suggests AI-generated content (more predictable)
• Statistical Modeling: Uses probability distributions to assess text complexity

3. Machine Learning Classification 🤖

Uses trained models to classify content:

• TF-IDF Vectorization: Converts text to numerical features
• Random Forest Classifier: Ensemble learning for robust predictions
• Synthetic Training Data: Generated human and AI text samples
• Probability Scoring: Provides confidence levels for predictions

4. Combined Analysis 🎯

Integrates all three methods for maximum accuracy:

• Weighted Combination: Balances different analysis methods
• Cross-Validation: Reduces false positives and negatives
• Robust Detection: Handles sophisticated AI-generated text

📊 Understanding Results

Confidence Levels

• High Confidence (>70%): Strong indication of content origin
• Medium Confidence (50-70%): Mixed signals, consider additional context
• Low Confidence (<50%): Uncertain results, manual review recommended

Visual Indicators

• Pie Charts: Show probability distribution
• Gauge Charts: Display perplexity scores
• Bar Charts: Feature breakdown analysis
• Color Coding: Green for human, red for AI indicators

Detailed Metrics

• Human-Written Probability: Percentage indicating human authorship
• AI-Generated Probability: Percentage indicating AI generation
• Confidence Level: Overall reliability of the analysis
• Feature Scores: Individual characteristic measurements

🛠️ Technical Details

Architecture

• Frontend: Streamlit web application
• Backend: Python-based analysis engine
• ML Pipeline: Scikit-learn for classification
• Text Processing: NLTK for natural language processing
• Visualization: Plotly for interactive charts

Dependencies

• streamlit: Web application framework
• numpy: Numerical computing
• pandas: Data manipulation
• scikit-learn: Machine learning algorithms
• plotly: Interactive visualizations
• nltk: Natural language processing
• python-docx: Word document processing
• PyPDF2: PDF text extraction

Model Training

The system automatically trains a machine learning model using:

• Synthetic Human Text: Generated samples with natural variations
• Synthetic AI Text: Generated samples with AI-like patterns
• Feature Engineering: TF-IDF vectorization of text
• Model Persistence: Saves trained models for reuse

🔧 Configuration

Analysis Settings

• Detection Method: Choose analysis approach
• Detailed Explanation: Toggle detailed feature breakdown
• Feature Breakdown: Show individual feature scores

Model Management

• Automatic Training: Models train on first run
• Model Persistence: Trained models are saved locally
• Model Status: Real-time model availability indicators

📈 Performance

Accuracy

• Combined Analysis: Highest accuracy across all methods
• Cross-Validation: Robust against different text types
• False Positive Reduction: Minimizes incorrect AI detections

Speed

• Real-time Analysis: Instant results for most text lengths
• Optimized Processing: Efficient algorithms for large documents
• Caching: Model persistence for faster subsequent runs

🤝 Contributing

We welcome contributions! Here's how you can help:

Development Setup

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

Areas for Improvement

• Enhanced ML Models: Better training data and algorithms
• Additional Features: More detection methods
• UI Improvements: Better user experience
• Performance Optimization: Faster processing
• Documentation: More detailed guides

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Streamlit: For the excellent web application framework
• Scikit-learn: For machine learning capabilities
• NLTK: For natural language processing tools
• Plotly: For interactive visualizations
• Open Source Community: For the libraries that make this possible

📞 Support

If you encounter any issues or have questions:

• GitHub Issues: Report bugs and feature requests
• Documentation: Check this README for usage instructions
• Community: Join discussions in the repository

---

🔍 AI Content Detector Pro - Advanced AI-generated content detection using multiple analysis techniques