bee-taxonomy 0.0.7

A library for taxonomy management.

Taxonomy Library README

This module provides functions for taxonomy creation, text classification, and header translation using AI models and vector similarity search.

Usage

Import the module with:

from bee_taxonomy import taxonomy

Main Functions

1. taxonomy.propose_taxonomy(field: str, description: str, discrete_fields: list[str] = None)

Purpose: Generate taxonomy suggestions using OpenAI Parameters:

• field: Name of the field to categorize
• description: Description of the field's purpose
• discrete_fields: Optional specific values to consider

Example:

taxonomy.propose_taxonomy(
    field="Color",
    description="Vehicle paint color classification",
    discrete_fields=["Red", "Blue", "Green", "Custom"]
)
# Returns: ["Red", "Blue", "Green", "Other"]

2. taxonomy.apply_taxonomy_similarity(discrete_fields: list[str], taxonomy: list[str], category_type: str = None)

Purpose: Classify values using semantic similarity with vector database Parameters:

• discrete_fields: Values to classify
• taxonomy: List of allowed classification terms
• category_type: Special processing for categories like 'streets'

Example:

taxonomy.apply_taxonomy_similarity(
    discrete_fields=["Rd", "Street", "Ave"],
    taxonomy=["Road", "Street", "Avenue"],
    category_type="streets"
)
# Returns: {'Rd': {'match': 'Road', 'score': 0.92}, ...}

3. taxonomy.apply_taxonomy_reasoning(discrete_fields: list[str], taxonomy: list[str], classification_description: str, hash_file: str = None)

Parameters:

• discrete_fields: List of values to classify
• taxonomy: List of allowed categories
• classification_description: Context for classification
• hash_file: Optional file hash for progress tracking

Example:

taxonomy.apply_taxonomy_reasoning(
    discrete_fields=["Quick Brown Fox", "Lazy Dog"],
    taxonomy=["Animal", "Object", "Action"],
    classification_description="Classify animal-related phrases"
)
# Returns: {'Quick Brown Fox': 'Animal', 'Lazy Dog': 'Animal'

4. taxonomy.translate_headers_reasoning(src_lang, dest_lang, headers)

Parameters:

• src_lang: Source language code
• dest_lang: Target language code
• headers: List of headers to translate

Example:

taxonomy.translate_headers_reasoning(
    src_lang="en",
    dest_lang="es",
    headers=["Street Name", "Zip Code"]
)
# Returns: {'Street Name': 'Nombre de la Calle', 'Zip Code': 'Código Postal'

5. taxonomy.analyze_text_field(field_name: str, field_value: str, task: Literal["label", "summarize"] = "label")

Parameters:

• field_name: Name of the text field
• field_value: Text to analyze
• task: "label" for classification or "summarize" for text summary

Example:

taxonomy.analyze_text_field(
    field_name="Product Description",
    field_value="This ergonomic chair provides lumbar support and adjustable height",
    task="label"
)
# Returns: "Office Furniture"

Environment Variables

• MODEL_NAME: Hugging Face model identifier
• SERVER_URL: Base URL for OpenAI-compatible API
• API_KEY: Authentication token for the API
• EMBEDDER_MODEL: Embedding model for semantic similarity

Features

• Validation workflow with Pydantic models
• Progress checkpointing for large datasets
• Google search integration for ambiguous classifications