Product Taxonomy Mapper API

Map product catalogs to your own category tree. The Actor accepts pasted JSON or an Apify dataset, then returns a category, confidence score, matched evidence, alternatives, and a review flag for every product.

🎯 What it does

• Maps products to any merchant, marketplace, or internal taxonomy.
• Uses product titles, descriptions, brands, product types, category keywords, and examples.
• Keeps uncertain results visible with needsReview instead of inventing certainty.
• Works without browser scraping, proxies, or a third-party AI key.

🚀 Quick start

Provide products and taxonomy, then click Start. The prefilled input is a complete working example.

{
  "products": [{"id":"sku-1","title":"Women's waterproof hiking boots"}],
  "taxonomy": [
    {"id":"boots","name":"Hiking Boots","path":"Apparel > Footwear > Boots","keywords":["hiking","trail","waterproof"]},
    {"id":"sneakers","name":"Sneakers","keywords":["trainer","running"]}
  ]
}

📥 Input

Use products for direct JSON rows or datasetId to chain another Actor's dataset. taxonomy is required. Tune minimumConfidence to control the manual-review threshold and maxItems to cap a run.

📤 Output

Each source row is preserved and enriched with mappedCategoryId, mappedCategoryName, mappedCategoryPath, confidence, needsReview, matchedSignals, and three ranked alternatives.

🧠 How scoring works

The mapper normalizes Unicode and combines exact category phrases, category/path token overlap, weighted keywords, and example overlap. Confidence also considers the margin between the best and second-best category. The method is deterministic and auditable.

💵 Pricing

The intended Store price is $2 per 1,000 mapped product rows, including platform usage. One dataset row is one billable result.

🔗 Common workflows

• Map Shopify products to a marketplace taxonomy.
• Normalize supplier catalogs before Google Merchant export.
• Route 1688 sourcing data into an internal category tree.
• Flag low-confidence catalog rows for human review.

⚠️ Limits

Quality depends on category names, keywords, examples, and product text. Highly ambiguous products should remain reviewable. The Actor does not claim semantic understanding beyond the evidence it returns.

🛟 Support

Open an issue on the Actor page with a minimal redacted input, expected category, and actual output.