AI Category Matching

AI Category Matching - Intelligent Product Categorization

Overview

AI-powered product categorization using Claude's natural language understanding to automatically assign products to the most appropriate categories based on product names, descriptions, and attributes.

Configuration

• App ID: df7c18eb-009b-4868-9891-15e19dddbe67
• API Key: ${API_KEY}
• Site ID: ${SITE_ID}

How AI Category Matching Works

Step 1: Analyze Product

Claude analyzes multiple product attributes:

Primary Signals:

• Product name (strongest signal)
• Product description
• Product type (physical/digital/service)
• Price range
• Existing tags/keywords

Secondary Signals:

• SKU patterns
• Brand information
• Product images (if using vision)
• Similar products' categories

Step 2: Analyze Available Categories

For each category in the store:

• Extract category name and description
• Identify category keywords
• Analyze existing products in category
• Understand category theme/purpose

Step 3: Match Products to Categories

Matching Algorithm:

For each product:
  1. Extract keywords from name + description
  2. For each category:
     a. Calculate keyword overlap
     b. Check semantic similarity
     c. Consider category product patterns
     d. Generate confidence score
  3. Rank categories by score
  4. Return top match with confidence level

Confidence Levels:

• High (90-100%): Clear match, highly confident
• Medium (70-89%): Good match, some ambiguity
• Low (50-69%): Possible match, needs review
• Uncertain (<50%): No clear match, manual review needed

Example Matching Scenarios

Scenario 1: Clear Match

Product: "Wireless Bluetooth Headphones - Premium Sound"

Analysis:

Keywords: wireless, bluetooth, headphones, audio, sound, premium
Product Type: Electronics
Price: $89.99 (mid-to-high range)

Category Matching:
1. Electronics → 95% (keywords: bluetooth, wireless, tech)
2. Audio & Music → 88% (keywords: headphones, sound)
3. Accessories → 45% (generic match)

✅ Best Match: Electronics (95% confidence - HIGH)

Scenario 2: Multi-Category Product

Product: "Yoga Mat - Eco-Friendly Exercise Mat with Carrying Strap"

Analysis:

Keywords: yoga, mat, exercise, fitness, eco-friendly, workout
Product Type: Physical
Price: $45.00

Category Matching:
1. Sports & Fitness → 92% (yoga, exercise, fitness)
2. Yoga & Meditation → 90% (yoga, mat)
3. Eco-Friendly Products → 75% (eco-friendly)

✅ Best Match: Sports & Fitness (92% confidence - HIGH)
💡 Suggestion: Also add to "Yoga & Meditation" and "Eco-Friendly Products"

Scenario 3: Ambiguous Product

Product: "Gift Set - Assorted Items"

Analysis:

Keywords: gift, set, assorted
Product Type: Physical
Price: $50.00

Category Matching:
1. Gifts & Bundles → 65% (gift keyword)
2. Special Offers → 45% (set, bundle)
3. Uncategorized → 40%

⚠️ Best Match: Gifts & Bundles (65% confidence - MEDIUM)
💡 Recommendation: Add more specific description or create custom category

Workflow: Auto-Categorize All Products

Full Automation Sequence

SITE_ID="${SITE_ID}"
API_KEY="${API_KEY}"

echo "Step 1: Fetching all categories..."
categories=$(curl -s -X POST "https://www.wixapis.com/stores/v1/collections/query" \
  -H "Authorization: ${API_KEY}" \
  -H "wix-site-id: ${SITE_ID}" \
  -H "Content-Type: application/json" \
  -d '{"query": {"paging": {"limit": 100}}}')

echo "Found $(echo "$categories" | jq '.collections | length') categories"

echo "\nStep 2: Fetching all products..."
products=$(curl -s -X POST "https://www.wixapis.com/stores/v1/products/query" \
  -H "Authorization: ${API_KEY}" \
  -H "wix-site-id: ${SITE_ID}" \
  -H "Content-Type: application/json" \
  -d '{"query": {"paging": {"limit": 100}}}')

echo "Found $(echo "$products" | jq '.products | length') products"

echo "\nStep 3: AI analyzing and matching..."

# For each product
echo "$products" | jq -r '.products[] | @json' | while read -r product_json; do
  product_id=$(echo "$product_json" | jq -r '.id')
  product_name=$(echo "$product_json" | jq -r '.name')
  product_desc=$(echo "$product_json" | jq -r '.description // ""')

  # Claude analyzes product against all categories
  # Returns best matching category ID and confidence
  # (This would be done via Claude's analysis capabilities)

  echo "  📦 ${product_name}"
  echo "     → Matched to: [CATEGORY_NAME] (confidence: XX%)"

  # Assign product to category
  # curl PATCH to update product.collectionIds
done

echo "\n✅ Auto-categorization complete!"

Bulk Category Creation + Smart Assignment

Create Category and Auto-Populate

Use Case: Create "Electronics" category and automatically add all electronic products

echo "Creating 'Electronics' category..."

# Step 1: Create category
category_response=$(curl -s -X POST "https://www.wixapis.com/stores/v1/collections" \
  -H "Authorization: ${API_KEY}" \
  -H "wix-site-id: ${SITE_ID}" \
  -H "Content-Type: application/json" \
  -d '{
  "collection": {
    "name": "Electronics",
    "description": "Electronic devices, gadgets, and tech accessories",
    "visible": false
  }
}')

CATEGORY_ID=$(echo "$category_response" | jq -r '.collection.id')
echo "✅ Created category: $CATEGORY_ID"

# Step 2: Get all products
all_products=$(curl -s -X POST "https://www.wixapis.com/stores/v1/products/query" \
  -H "Authorization: ${API_KEY}" \
  -H "wix-site-id: ${SITE_ID}" \
  -H "Content-Type: application/json" \
  -d '{"query": {"paging": {"limit": 100}}}')

# Step 3: AI filters products that match "Electronics"
# Claude identifies products with keywords: phone, computer, gadget, headphone, charger, cable, etc.

echo "\nStep 3: AI analyzing products for Electronics match..."

# Step 4: Bulk assign matched products
matched_count=0

echo "$all_products" | jq -r '.products[] | select(.name | test("phone|computer|headphone|gadget|charger|cable|tech|electronic"; "i")) | .id' | while read -r pid; do
  curl -s -X PATCH "https://www.wixapis.com/stores/v1/products/${pid}" \
    -H "Authorization: ${API_KEY}" \
    -H "wix-site-id: ${SITE_ID}" \
    -H "Content-Type: application/json" \
    -d '{
    "product": {
      "id": "'"${pid}"'",
      "collectionIds": ["'"${CATEGORY_ID}"'"]
    }
  }' > /dev/null

  matched_count=$((matched_count + 1))
  echo "  ✅ Added product to Electronics"
  sleep 0.2
done

echo "\n✅ Added ${matched_count} products to Electronics category"

# Step 5: Make category visible
curl -s -X PATCH "https://www.wixapis.com/stores/v1/collections/${CATEGORY_ID}" \
  -H "Authorization: ${API_KEY}" \
  -H "wix-site-id: ${SITE_ID}" \
  -H "Content-Type: application/json" \
  -d '{
  "collection": {
    "visible": true
  }
}' > /dev/null

echo "✅ Category is now visible with all matched products!"

Category Optimization Strategies

1. Consolidate Similar Categories

If you have: "Tech", "Electronics", "Gadgets" → Merge into single "Electronics"

2. Create Missing Categories

AI can identify underserved product groups:

• Products with no category
• Products in "Uncategorized"
• Clusters of similar products without a home

3. Category Performance Analysis

Track which categories drive sales:

# Get products by category with revenue data
# Identify high-performing categories
# Suggest creating similar categories

Advanced Features

Multi-Category Assignment

Products can belong to multiple categories:

{
  "collectionIds": [
    "electronics",
    "sale-items",
    "new-arrivals",
    "bestsellers"
  ]
}

Recommended Multi-Category Patterns:

• Primary category (product type)
• Secondary categories (characteristics):
  • Sale Items
  • New Arrivals
  • Bestsellers
  • Seasonal (Summer Collection, Holiday Gifts)
  • Special (Eco-Friendly, Handmade, Local)

Category Hierarchy

Create nested categories:

Electronics
  ├─ Phones & Tablets
  ├─ Computers
  ├─ Audio & Video
  └─ Accessories

Use parentId field in category creation.