Emotional-Aware Chat System

A Rust-based conversational AI that detects user emotions and dynamically adjusts response strategies using Zhipu AI's GLM-4.7 API.

Features

• 🎭 Emotion Detection - Analyzes user sentiment in real-time (Positive/Negative/Neutral)
• 📈 Emotion Trend Tracking - Monitors how emotions change over the conversation
• 🎯 Dynamic Response Strategies - Adapts conversation style based on emotional state:
  • Empathetic - For users in distress (negative + declining trend)
  • Encouraging - For users needing motivation (negative + stable trend)
  • Cheerful - For users in good mood (positive sentiment)
  • Neutral - Professional, balanced responses (default)
• 💬 Context-Aware - Maintains conversation history for coherent multi-turn dialogue
• 🛡️ Error Handling - Graceful fallback for API failures and edge cases

Architecture

User Input
    ↓
EmotionDetector → SentimentClassification
    ↓
ConversationManager → EmotionTrend
    ↓
select_strategy() → ResponseStrategy
    ↓
ChatAgent → Context-Aware Response
    ↓
Output (emotion, trend, strategy, response)

Tech Stack

• Language: Rust 2024 Edition
• AI Framework: rig-core 0.11.1
• Runtime: tokio 1.34
• Serialization: serde, schemars
• Model: Zhipu AI GLM-4.7

Prerequisites

• Rust 1.93+
• Zhipu AI API key (Get one here)

Installation

# Clone the repository
git clone https://github.com/Clearzero22/text_classifier_extractor.git
cd text_classifier_extractor

# Create environment file
cp .env.example .env

# Edit .env with your API key
# OPENAI_API_KEY=your-api-key-here

Configuration

Create a .env file in the project root:

# Required
OPENAI_API_KEY=your-zhipu-ai-api-key

# Optional (with defaults shown)
OPENAI_BASE_URL=https://open.bigmodel.cn/api/paas/v4
MODEL=glm-4.7

Usage

# Run the application
cargo run --release

# Or in development mode
cargo run

Example Session

🤖 Emotional-Aware Chat System
📊 Model: glm-4.7
💬 Type 'quit' or 'exit' to end

You: Hello! How are you today?
📊 Emotion: Positive (confidence: 0.95)
📈 Trend: Stable
🎯 Strategy: Cheerful
🤖 Assistant: Hello! I'm doing wonderfully, thank you for asking! 😊
It's such a pleasure to chat with someone so friendly. How can I brighten your day?

You: I'm feeling really down lately
📊 Emotion: Negative (confidence: 0.85)
📈 Trend: Declining
🎯 Strategy: Empathetic
🤖 Assistant: I hear you, and I want you to know that your feelings are completely valid.
It takes courage to share when you're struggling. Please know that I'm here to listen without judgment.

You: quit
👋 Goodbye!

Development

Running Tests

# Run all tests
cargo test

# Run specific test module
cargo test emotion
cargo test conversation
cargo test strategy

Project Structure

src/
├── main.rs              # Entry point, CLI interface
├── models/
│   └── message.rs       # Message and MessageRole types
├── agents/
│   ├── emotion.rs       # EmotionDetector using structured extraction
│   └── chat.rs          # ChatAgent with strategy-based responses
├── state/
│   └── conversation.rs  # ConversationManager, EmotionTrend
└── strategy/
    └── response.rs      # ResponseStrategy enum and selection logic

API Integration

This project uses Zhipu AI's GLM-4.7 model through an OpenAI-compatible API:

• Provider: Zhipu AI (BigModel)
• Documentation: https://open.bigmodel.cn/
• Structured Extraction: Uses JSON schema for type-safe sentiment analysis

Error Handling

The system includes robust error handling:

• Short Text Handling: Enhanced prompts for inputs < 5 characters
• API Failures: Automatic fallback to Neutral sentiment
• Edge Cases: Handles empty history, single emotion, and boundary conditions

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is open source and available under the MIT License.

Acknowledgments

• Built with rig-core - LLM framework for Rust
• Powered by Zhipu AI GLM-4.7