4D Demo AI Similarities

A demonstration project showcasing how to use AI-powered vector embeddings to find similar customers in a 4D database application. This project leverages the 4D AI Kit to detect potential duplicate customers based on semantic similarity rather than exact matches.

🎯 Overview

This application demonstrates how to use artificial intelligence to identify similar customers in a database by:

• Generating AI embeddings for customer data
• Computing vector similarities to find duplicates
• Providing an intuitive interface for managing customer data and similarity searches

✨ Features

Customer Management

• AI-Powered Customer Generation: Automatically generate realistic customer data using AI
• Customer Creation Form: Manual customer entry
• Smart Vectorization: Convert customer information into AI embeddings for similarity comparison

Similarity Detection

• Vector-Based Matching: Uses cosine similarity to find potential duplicates
• Configurable Similarity Threshold: Adjust the sensitivity of similarity detection
• Batch Processing: Search similarities across the entire customer database
• Real-time Results: Fast similarity calculations with performance metrics

AI Provider Support

• OpenAI Integration: Support for OpenAI's embedding models (for instance text-embedding-ada-002)
• Ollama Support: Local AI model support via Ollama (for instance nomic-embed-text)
• Flexible Configuration: Easy switching between different AI providers and models

🚀 Getting Started

Prerequisites

• 4D 20 R10 or later
• 4D AI Kit component
• AI Provider access:
  • OpenAI: API key for OpenAI services
  • Ollama: Local or remote Ollama installation with embedding models

Installation

1. Clone this repository
2. Open the project in 4D: 4D_Demo_AI_Similarities.4DProject
3. Run the application
4. Configure your AI provider settings in the application

📋 Usage

1. Generate Sample Data

• Select your AI provider and embedding model
• Use the AI Customer Generator to create realistic test customers
• The system generates complete customer profiles with names, emails, phones, and addresses
• Monitor progress as customers are generated

2. Create Vector Embeddings

• Select your AI provider and embedding model
• Click "Generate Embeddings" to vectorize all customer data
• Monitor progress as embeddings are generated

3. Search for Similarities

• Go to the Search Similarities form
• Adjust the similarity threshold (recommended: 90-95%)
• Click "Search similarities" to find potential duplicates
• Review results with similarity scores and customer details

4. Manual Customer Entry

• Use the Create Customer form for manual data entry
• New customers are automatically vectorized if embeddings are configured
• Search for similarities before saving

🏗️ Technical Architecture

Core Classes

• AI_Agent: Base class for AI operations with provider management
• AI_Vectorizer: Handles text-to-vector conversion using AI embeddings
• AI_CustomerGenerator: Generates realistic customer data using AI
• customerEntity: Customer data model with vectorization capabilities
• customer: DataClass with similarity search algorithms

Database Structure

• Customer Table: Stores customer information and vector embeddings
• Provider Settings: AI provider configurations and API keys
• Embedding Info: Tracks vectorization status and metadata

🤝 Contributing

This is a demonstration project showcasing 4D AI Kit capabilities. Feel free to:

• Fork the repository for your own experiments
• Submit issues for bugs or feature requests
• Share improvements and optimizations

📄 License

This project is provided as-is for educational and demonstration purposes.

🙏 Acknowledgments

• Built with 4D database platform
• Powered by 4D AI Kit
• Uses OpenAI and Ollama for AI embeddings and generative AI
• Demonstrates modern AI integration in database applications

---

Note: This project requires valid AI provider credentials and is intended for demonstration purposes. Production use should include additional error handling, security measures, and performance optimizations.