Concepts

Hierarchical Taxonomy Classification

Assigns content to multi-level taxonomies using embedding-based classification. Taxonomies are reusable control planes that define organizational structure.

video

image

text

Multi-Tier

76.0K runs

Deploy Recipe

Why This Matters

Taxonomies are infrastructure—not model outputs. Once defined, they enable consistent classification, compliance tagging, and structured filtering across all content.

from mixpeek import Mixpeek

client = Mixpeek(api_key="your-api-key")

# Define taxonomy structure
taxonomy = client.taxonomies.create(
    taxonomy_name="content_classification",
    hierarchy={
        "safe_content": {
            "educational": ["tutorial", "documentary"],
            "entertainment": ["comedy", "music"]
        },
        "review_required": {
            "ambiguous": ["political", "news"]
        }
    }
)

# Classify content
result = client.collections.classify(
    collection_id="my-collection",
    taxonomy_id=taxonomy.id,
    confidence_threshold=0.75
)

# Filter by taxonomy in retriever
results = client.retrievers.execute(
    retriever_id="filtered-retriever",
    inputs={
        "query_text": "educational videos",
        "taxonomy_path": "safe_content.educational"
    }
)

Retrieval Flow

attribute filter

Filter by taxonomy labels

sort

Sort by confidence score

attribute filter(filter)

Filter by taxonomy labels

sort(rank)

Sort by confidence score

Feature Extractors

Image Embedding Text Embedding Video Embedding

Feature Extractors

Image Embedding

Generate visual embeddings for similarity search and clustering

752K runs

Text Embedding

Extract semantic embeddings from documents, transcripts and text content

827K runs

Video Embedding

Generate vector embeddings for video content

610K runs

Retriever Stages

attribute filter

Filter documents by metadata attributes

filter

sort

Sort documents by field values

rank

Enrichment Resources

Taxonomy

Documentation

Taxonomies