Semantic Join

import requests

API_URL = "https://api.mixpeek.com"
headers = {"Authorization": "Bearer YOUR_API_KEY", "X-Namespace": "your-namespace"}

# Create taxonomy to join video content with product catalog
taxonomy = requests.post(f"{API_URL}/v1/taxonomies", headers=headers, json={
    "taxonomy_name": "product_matcher",
    "taxonomy_type": "flat",
    "retriever_id": "ret_product_search",
    "input_mappings": {
        "query_embedding": "mixpeek://multimodal_extractor@v1/embedding"
    },
    "source_collection": {
        "collection_id": "col_product_catalog",
        "enrichment_fields": [
            {"field_path": "metadata.sku", "merge_mode": "enrich"},
            {"field_path": "metadata.category", "merge_mode": "enrich"}
        ]
    }
}).json()

# Apply taxonomy to video collection (semantic join)
requests.post(
    f"{API_URL}/v1/collections/col_marketing_videos/apply-taxonomy",
    headers=headers,
    json={"taxonomy_id": taxonomy["taxonomy_id"]}
)

# Search videos - results now include matched product data
results = requests.post(
    f"{API_URL}/v1/retrievers/video-search/execute",
    headers=headers,
    json={"query": {"text": "product demos"}}
).json()

for doc in results["documents"]:
    print(f"Video: {doc['document_id']}")
    print(f"  Matched SKU: {doc.get('metadata.sku', 'N/A')}")
    print(f"  Category: {doc.get('metadata.category', 'N/A')}")