liveBook · Manning

Similarity search

As explained in AI Governance

Similarity search is a retrieval technique that finds data items whose representations are close to a query representation in a vector space.

Similarity search retrieves records near a query in embedding space. In privacy-sensitive RAG systems, that retrieval can help link embeddings to individuals or surface other people’s data among nearby results. section 5.3.2 section 5.5.2

Treat retrieval as a privacy control point:

  • Check results for other people’s data before disclosing them in response to an access request. section 5.5.2
  • Ensure a hosted vector database supports exporting records per user; otherwise, access requests may be difficult to fulfill. section 5.5.2

How AI Governance covers it

How deep it goes
Connects retrieval to re-identification and cross-person disclosure risks, but gives no retrieval algorithm or implementation detail.
Examples
Uses RAG-indexed emails and nearby records belonging to other people as disclosure-risk examples.
What you can do after
Apply disclosure checks to retrieved neighbors and verify that vendors can export records per user.
How it connects
Links similarity search with embeddings, RAG, personal-data rights, and vector databases.

Also explained in

Compare how 37 books explain similarity search →

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.