Emakia Node

Advanced content analysis platform with multiple AI agents for detecting toxicity, bias, misinformation, and coordination patterns in text content.

Features

• Real AI Models: Uses Cerebras Qwen-3-32B for intelligent analysis
• Multi-Agent Analysis: Four specialized AI agents working together
• Toxicity Detection: AI-powered detection of harmful and inappropriate content
• Bias Detection: Intelligent detection of bias and discrimination patterns
• Misinformation Detection: AI analysis of false claims and misleading content
• Coordination Detection: Detection of coordinated behavior and manipulation
• Modern Web UI: Responsive interface with real-time AI analysis
• RESTful API: Comprehensive API for integration with other systems
• Context Understanding: Real AI models understand meaning, not just keywords

Project Structure

emakia-node/
├── agents/
│   ├── toxicityAgent.js      # Toxicity detection agent
│   ├── biasAgent.js          # Bias detection agent
│   ├── misinformationAgent.js # Misinformation detection agent
│   └── coordinationAgent.js   # Coordination detection agent
├── routes/
│   └── classify.js           # API routes for classification
├── public/
│   └── index.html            # Web interface
├── app.js                    # Main server file
├── package.json              # Dependencies and scripts
└── README.md                # This file

Prerequisites

• Node.js (version 16 or higher)
• npm or yarn package manager

Installation

1. Clone or navigate to the project directory:

   cd emakia-node

2. Install dependencies:

   npm install

3. Set up environment variables (optional for enhanced AI models):

   # Create .env file
   touch .env
   
   # Add your Cerebras API configuration (optional)
   echo "CEREBRAS_API_KEY=your_cerebras_api_key_here" >> .env
   echo "CEREBRAS_API_URL=https://api.cerebras.com" >> .env
   echo "CEREBRAS_MODEL=Qwen-3-32B" >> .env

   Note: The system works without the API key using intelligent fallback detection, but the Cerebras Qwen-3-32B model provides superior accuracy for complex cases.

Usage

Development Mode

npm run dev

Production Mode

npm start

The application will start on http://localhost:3000

API Endpoints

POST /api/classify

Analyze text with all or selected agents.

Request Body:

{
  "text": "Your text to analyze",
  "agents": ["toxicity", "bias", "misinformation", "coordination"]
}

Response:

{
  "text": "Your text to analyze",
  "timestamp": "2024-01-01T12:00:00.000Z",
  "analyses": {
    "toxicity": {
      "isToxic": true,
      "toxicityScore": 0.8,
      "severity": "high",
      "categories": {...},
      "recommendations": [...]
    },
    "bias": {...},
    "misinformation": {...},
    "coordination": {...}
  },
  "summary": {
    "totalIssues": 3,
    "criticalIssues": 1,
    "highIssues": 2,
    "mediumIssues": 0,
    "lowIssues": 0
  },
  "overallRisk": "critical"
}

GET /api/classify

Analyze text using query parameters.

Example:

GET /api/classify?text=your message&agents=toxicity,bias

Individual Agent Endpoints

• POST /api/classify/toxicity - Toxicity analysis only
• POST /api/classify/bias - Bias analysis only
• POST /api/classify/misinformation - Misinformation analysis only
• POST /api/classify/coordination - Coordination analysis only

GET /api/health

Check system health and agent status.

GET /api/agents

Get information about available agents.

Agent Capabilities

ToxicityAgent

• Categories: Hate speech, violence, harassment, profanity
• Severity Levels: Low, medium, high, critical
• Features: Pattern matching, severity assessment, recommendations

BiasAgent

• Categories: Gender, racial, age, religious, socioeconomic bias
• Indicators: Absolute statements, generalizations, stereotypes, loaded language
• Features: Bias detection, inclusive language recommendations

MisinformationAgent

• Categories: Conspiracy theories, pseudoscience, clickbait, sensationalism
• Red Flags: Emotional language, urgency, authority appeals, exclusivity
• Features: Fact-checking guidance, source verification

CoordinationAgent

• Categories: Bot-like behavior, coordinated campaigns, manipulation tactics
• Indicators: Urgency pressure, emotional manipulation, social proof
• Features: Authenticity verification, manipulation detection

Configuration

Environment Variables

Create a .env file with the following variables:

# Server Configuration
PORT=3000
NODE_ENV=development

# API Configuration
API_VERSION=v1
API_TIMEOUT=30000

# Agent Configuration
AGENT_TIMEOUT=10000
AGENT_CONFIDENCE_THRESHOLD=0.6

# Logging Configuration
LOG_LEVEL=info
LOG_FORMAT=combined

# Security Configuration
CORS_ORIGIN=*
RATE_LIMIT_WINDOW=15
RATE_LIMIT_MAX=100

Agent Customization

Each agent can be customized by updating patterns and thresholds:

// Example: Update toxicity patterns
const toxicityAgent = new ToxicityAgent();
toxicityAgent.updatePatterns({
  new_category: ['new', 'keywords', 'here']
});

// Example: Update severity thresholds
toxicityAgent.updateSeverityLevels({
  custom: 0.5
});

Web Interface

The web interface provides:

• Multi-agent selection: Choose which agents to use
• Real-time analysis: Instant results with visual indicators
• Detailed breakdown: Individual agent results with scores
• Recommendations: Actionable advice for each detected issue
• Responsive design: Works on desktop and mobile devices

Development

Project Structure Details

• app.js: Express server with agent initialization and route handling
• agents/: Individual agent classes with specialized analysis logic
• routes/classify.js: API endpoints for text classification
• public/index.html: Modern web interface with JavaScript

Adding New Agents

1. Create a new agent class in agents/
2. Implement the required methods: analyze(), getName(), getDescription(), etc.
3. Add the agent to app.js
4. Update the web interface to include the new agent

Extending Existing Agents

Each agent supports:

• Pattern updates via updatePatterns()
• Threshold updates via updateSeverityLevels()
• Custom analysis logic in the analyze() method

Troubleshooting

Common Issues

1. Port already in use:

   PORT=3001 npm start

2. Agent errors:

   • Check agent initialization in app.js
   • Verify agent class exports
   • Review console logs for specific errors

3. API errors:

   • Verify request format
   • Check required fields (text, agents)
   • Review server logs

Debug Mode

Enable debug logging:

LOG_LEVEL=debug npm start

Security Considerations

• Input validation on all endpoints
• Rate limiting to prevent abuse
• Error handling without exposing sensitive information
• CORS configuration for cross-origin requests

Performance

• Agent analysis runs asynchronously
• Configurable timeouts for external API calls
• Efficient pattern matching algorithms
• Memory-conscious text processing

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test thoroughly
5. Submit a pull request

License

MIT License - see LICENSE file for details.

Support

For issues and questions:

• Check the troubleshooting section
• Review the API documentation
• Open an issue on GitHub

---

Note: This is a demonstration platform. For production use, implement proper security measures, error handling, and integrate with real AI models for improved accuracy.