Emotion-Aware Interactive Storytelling

An AI-powered interactive storytelling system that adapts narratives based on emotional analysis. This project combines natural language processing for emotion detection with context-aware story generation to create dynamic, emotionally responsive gaming experiences.

Core Features

• 🎮 Interactive text-based storytelling with emotional awareness
• 🧠 Advanced emotion analysis using DistilRoBERTa model
• 🔄 Context-aware emotion transformation system
• 📊 Real-time emotional state tracking
• 🌍 Dynamic context adaptation
• 📈 Performance metrics and testing framework

Technical Implementation

Emotion Analysis System

• Base emotion detection using j-hartmann/emotion-english-distilroberta-base
• 7 primary emotions: anger, disgust, fear, joy, neutral, sadness, surprise
• Intensity scoring (0.0-1.0) for emotion strength
• Context-aware emotion transformation based on narrative situation

Context Processing

• Semantic analysis using sentence-transformers/all-MiniLM-L6-v2
• Multiple context types: combat, danger, discovery, social, etc.
• Narrative tones: positive, negative, tense, mysterious, etc.
• Dynamic intensity adjustment based on context

Prerequisites

• Python 3.8 or higher
• 8GB RAM minimum (16GB recommended)
• Git for version control
• Basic understanding of NLP concepts

Installation

1. Clone the repository:

   git clone <repository-url>
   cd emotion_storytelling

2. Create and activate a virtual environment:

   # On macOS/Linux
   python3 -m venv venv
   source venv/bin/activate
   
   # On Windows
   python -m venv venv
   venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

Running the Application

1. Ensure your virtual environment is activated
2. Launch the Streamlit interface:

   streamlit run app.py

3. Access the application at http://localhost:8501

Project Structure

emotion_storytelling/
├── app.py                 # Main Streamlit application
├── emotion_analyzer.py    # Emotion analysis engine
├── context_feature_extractor.py  # Context processing
├── game_state.py         # State management
├── test_emotion_accuracy.py      # Testing framework
├── test_data.json        # Test cases
└── requirements.txt      # Dependencies

Performance Metrics

Current system performance (as of latest testing):

• Base Model: 81.0% accuracy, MAE = 0.206
• Context-Aware: 76.2% accuracy, MAE = 0.170
• Balanced test set across 7 emotions
• Context adaptation shows 17.4% MAE reduction

Development Status

Implemented Features

• ✅ Base emotion detection
• ✅ Context-aware transformation
• ✅ Testing framework
• ✅ Performance metrics
• ✅ Real-time analysis

Troubleshooting

Common issues and solutions:

1. Model Loading Errors: Ensure sufficient RAM and proper internet connection for initial model download
2. Accuracy Issues: Check context settings and transformation rules
3. Performance Problems: Monitor system resources and consider GPU acceleration

Acknowledgments

• Emotion analysis model: j-hartmann/emotion-english-distilroberta-base
• Semantic analysis: sentence-transformers/all-MiniLM-L6-v2
• Testing methodology based on standard NLP evaluation metrics