Multi-Structural Embeddings
This tutorial demonstrates how Corvic AI enables you to create embeddings from multiple data structures, including structured, unstructured, relational, and visual data. Learn how to generate high-quality embeddings that preserve the structure and context of your diverse data sources.What You’ll Learn
- Creating embeddings from multiple data structures
- Working with structured, unstructured, relational, and visual data
- Preserving data structure and context in embeddings
- Generating high-quality multi-structural embeddings
Key Concepts
Multi-Structural Data
Corvic AI supports creating embeddings from multiple data structures simultaneously, including structured data (tables, databases), unstructured data (text documents, PDFs), relational data (graph structures, relationships), and visual data (images, charts). This allows you to capture the full context and relationships within your data.Structure Preservation
Unlike traditional approaches that flatten everything into text, multi-structural embeddings preserve the original structure, relationships, and context of your data. This results in more accurate and meaningful embeddings that better represent your enterprise data.Creating Multi-Structural Embeddings
Step 1: Prepare Your Data Sources
Gather data from multiple sources including structured tables, unstructured documents, relational databases, and visual content. Ensure your data is clean and well-organized for optimal embedding generation.Step 2: Create Corvic Tables
Define Corvic Tables that incorporate multiple data structures. Specify entities from different data types and configure how they relate to each other, preserving the structural relationships in your embeddings.Step 3: Generate Embeddings
Generate embeddings using algorithms optimized for multi-structural data. Corvic AI automatically handles the complexity of combining different data types while maintaining structure and context.Benefits
Multi-structural embeddings enable you to capture the full richness of your enterprise data, including relationships, context, and structure that would be lost in text-only approaches. This results in more accurate search, better recommendations, and more meaningful insights from your data.Best Practices
Organize your data sources logically and ensure consistent schemas across related data. Use Corvic Tables that appropriately combine different data structures, and validate that your embeddings preserve the important relationships and context from your source data.Related Documentation
Corvic Tables
Learn how to create Corvic Tables from multiple data structures for distributed processing.
Spaces
Generate embedding spaces from multi-structural Corvic Tables.
Data Sources
Understand how to manage diverse data sources.
Agents
Use multi-structural embeddings in your agents.