E-commerce Products

Tutorial 2: E-commerce Product Catalog

Build a product search system that combines semantic similarity with exact category matching.

Goal

Create a product catalog where: - Name + Description form composite semantic embedding - Category must match exactly (high weight) - Brand is considered but less important

Step 1: Create Configuration

Create config/products.yaml:

schema:
  name:
    type: text
    required: true
  description:
    type: text
    required: true
  category:
    type: text
    required: true
  brand: text
  price:
    type: number
    required: true

embeddings:
  # Name embedding
  name_vec:
    field: name
    model: tfidf

  # Description embedding
  desc_vec:
    field: description
    model: tfidf

  # Combined product text embedding
  product_vec:
    combine:
      - ref: name_vec
        weight: 0.4
      - ref: desc_vec
        weight: 0.6

similarities:
  # Semantic text similarity
  text_sim:
    embedding: product_vec

  # Category must match exactly
  category_sim:
    field: category
    metric: exact

  # Brand similarity helps
  brand_sim:
    field: brand
    metric: exact

  # Combined similarity
  overall:
    combine:
      - ref: text_sim
        weight: 0.6
      - ref: category_sim
        weight: 0.3
      - ref: brand_sim
        weight: 0.1

network:
  edges:
    similarity: overall
    min: 0.5

Step 2: Add Products

from src.network_rag import NetworkRAG

rag = (NetworkRAG.builder()
       .with_storage('products.db')
       .with_tfidf_embeddings()
       .from_config('config/products.yaml')
       .build())

products = [
    {
        'id': 'mbp14',
        'name': 'MacBook Pro 14"',
        'description': 'Powerful laptop with M3 Pro chip, perfect for development and creative work',
        'category': 'laptops',
        'brand': 'Apple',
        'price': 1999.00
    },
    {
        'id': 'mbp16',
        'name': 'MacBook Pro 16"',
        'description': 'Premium laptop with M3 Max chip, exceptional performance for professionals',
        'category': 'laptops',
        'brand': 'Apple',
        'price': 2499.00
    },
    {
        'id': 'xps15',
        'name': 'Dell XPS 15',
        'description': 'High-performance laptop for developers, great keyboard and screen',
        'category': 'laptops',
        'brand': 'Dell',
        'price': 1799.00
    },
    {
        'id': 'airpods',
        'name': 'AirPods Pro',
        'description': 'Wireless earbuds with active noise cancellation',
        'category': 'audio',
        'brand': 'Apple',
        'price': 249.00
    },
    {
        'id': 'ipad',
        'name': 'iPad Pro 12.9"',
        'description': 'Powerful tablet with M2 chip, excellent for creative work and entertainment',
        'category': 'tablets',
        'brand': 'Apple',
        'price': 1099.00
    }
]

for product in products:
    rag.add(product['id'], document=product)

rag.build_network()

Step 3: Search Products

# Search for laptops
print("=== Search: 'powerful laptop for development' ===")
results = rag.search('powerful laptop for development').top(5)

for i, result in enumerate(results, 1):
    doc = result.metadata
    print(f"\n{i}. {doc['name']} (score: {result.score:.3f})")
    print(f"   Price: ${doc['price']}")
    print(f"   Category: {doc['category']}")
    print(f"   Description: {doc['description'][:60]}...")

Expected output:

1. MacBook Pro 14" (score: 0.782)
   Price: $1999.0
   Category: laptops
   Description: Powerful laptop with M3 Pro chip, perfect for development...

2. Dell XPS 15 (score: 0.765)
   Price: $1799.0
   Category: laptops
   Description: High-performance laptop for developers, great keyboard...

3. MacBook Pro 16" (score: 0.743)
   Price: $2499.0
   Category: laptops
   Description: Premium laptop with M3 Max chip, exceptional performance...

Step 4: Category-Filtered Search

# Search only within laptops category
results = (rag.search('Apple products')
           .filter(category='laptops')
           .top(3))

for result in results:
    doc = result.metadata
    print(f"{doc['name']}: ${doc['price']}")

Step 5: Find Similar Products

# Find products similar to a specific product
similar = rag.get_neighbors('mbp14', k_hops=1)

print(f"\nProducts similar to MacBook Pro 14\":")
for node_id in similar:
    node = rag.storage.get_node(node_id)
    doc = node['metadata']
    print(f"  - {doc['name']} ({doc['category']})")

---