Document Classification Pipeline

from mixpeek import Mixpeek

client = Mixpeek(api_key="YOUR_API_KEY")

# Create taxonomy for document types
taxonomy = client.taxonomies.create(
    namespace_id="ns_your_namespace",
    name="document_types",
    taxonomy_type="hierarchical",
    hierarchy=[
        {"node_id": "invoice", "collection_id": "col_invoice_examples"},
        {"node_id": "contract", "collection_id": "col_contract_examples"},
        {"node_id": "report", "collection_id": "col_report_examples"},
        {"node_id": "form", "collection_id": "col_form_examples"},
        {"node_id": "correspondence", "collection_id": "col_letter_examples"},
    ]
)

# Create document collection with layout extraction
collection = client.collections.create(
    namespace_id="ns_your_namespace",
    name="incoming_documents",
    extractors=["document-graph-extractor", "text-extractor"]
)

# Apply taxonomy for automatic classification
client.collections.apply_taxonomy(
    collection_id="col_incoming_documents",
    taxonomy_id=taxonomy["taxonomy_id"]
)

# Upload documents for classification
client.buckets.upload(bucket_id="bkt_docs", url="s3://your-bucket/incoming/")

# Query classified documents
docs = client.documents.list(
    collection_id="col_incoming_documents",
    filters={"taxonomy_enrichment.category": "invoice"}
)
print(f"Found {len(docs['results'])} invoices")