liveBook · Manning

Data labeling scheme

As explained in AI Governance

A data labeling scheme is a defined set of categories for classifying information by sensitivity or handling requirements.

A data labeling scheme classifies documents by sensitivity or handling requirements before they enter AI systems for fine-tuning, retrieval, or routine queries. section 6.6.2

Make labels enforceable across the data flow:

  • Assign a label when a document or knowledge source is added. section 6.6.2
  • Require vendors to preserve labels in stored data and return them in query results; enforce handling rules in your retrieval pipeline. section 6.6.2

How AI Governance covers it

How deep it goes
Defines four practical categories and ties labeling to intake, vendor preservation, and retrieval-side enforcement.
Examples
Uses public, internal, confidential, and trade-secret labels as practical categories.
What you can do after
Apply labels at intake, require vendors to preserve and return them, and enforce handling rules in retrieval.
How it connects
Connects information security, fine-tuning, retrieval-augmented generation, and data-loss prevention.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of AI Governance. The section links point to the parts of the book it draws on.