AI Embeddings and Vectors

Overview

Embeddings are numerical representations of text, images, or other data. Use them for semantic search, similarity matching, and building AI applications.

Concepts

• Embeddings: Fixed-size vectors representing meaning
• Vector Databases: Specialized databases for vector search
• Similarity Search: Find semantically similar items
• Semantic Search: Search by meaning rather than keywords
• Recommendations: Suggest relevant items based on similarity

Generate Embeddings

npm install openai

import OpenAI from 'openai'

const openai = new OpenAI({
  apiKey: process.env.OPENAI_API_KEY
})

async function getEmbedding(text: string) {
  const response = await openai.embeddings.create({
    model: 'text-embedding-3-small',
    input: text
  })
  
  return response.data[0].embedding
}

// Usage
const embedding = await getEmbedding('Hello world')
console.log(embedding.length) // 1536 dimensions

Vector Database with Pinecone

import { Pinecone } from '@pinecone-database/pinecone'

const pc = new Pinecone({
  apiKey: process.env.PINECONE_API_KEY
})

const index = pc.index('documents')

// Store embedding
await index.upsert([{
  id: 'doc-1',
  values: embedding,
  metadata: { text: 'Hello world', source: 'greeting' }
}])

// Search similar items
const results = await index.query({
  vector: embedding,
  topK: 5,
  includeMetadata: true
})

console.log(results.matches) // Top 5 similar documents

Semantic Search Implementation

async function semanticSearch(query: string, topK: number = 5) {
  const queryEmbedding = await getEmbedding(query)
  
  const results = await index.query({
    vector: queryEmbedding,
    topK,
    includeMetadata: true
  })
  
  return results.matches.map(match => ({
    text: match.metadata?.text,
    score: match.score,
    source: match.metadata?.source
  }))
}

const results = await semanticSearch('What is machine learning?')

Embeddings enable semantic understanding without training custom AI models.