Embedding

A numerical representation of text, images, or other data as vectors, enabling AI to measure similarity and meaning mathematically.

Embeddings convert text, images, or other data into numerical vectors that capture semantic meaning, enabling mathematical comparison of concepts.

How It Works

Text → Neural network → Vector of numbers (e.g., 1536 dimensions)

Similar concepts end up close together in vector space:

• “cat” and “kitten” → nearby vectors
• “cat” and “democracy” → distant vectors

Applications

• Semantic search: Find relevant documents by meaning, not keywords
• Recommendations: “Similar items” based on embeddings
• Clustering: Group related content automatically
• RAG: Retrieve relevant context for LLM queries
• Anomaly detection: Find outliers in embedding space

Popular Models

• OpenAI: text-embedding-3-small/large
• Cohere: embed-english-v3
• Sentence Transformers: Open-source options
• Voyage AI: Specialized embeddings

Vector Databases

Store and search embeddings efficiently:

• Pinecone, Weaviate, Chroma, Qdrant, pgvector