Skip to content
BENIBIX Learn

Vector Index

AI🔎 Retrieval & SearchVector Databases

Also known as ANN Index, Vector Search Index

A map that finds close vectors fast, so you do not look at all of them.

Think of it like

Think of it like signs in a big shop. They send you to the right aisle, so you skip the rest.

Example

Jack's database has 5 million vectors. Without an index, each search reads all 5 million. With one, it reads a few thousand and still finds the right matches.

Why it matters

Without an index, vector search slows down with every row you add.

Definition

A vector index is a structure that lets a database find close vectors without checking them all. One kind sorts vectors into groups and searches only the groups nearest the query. Another kind links each vector to its neighbors, so a search can hop toward the answer. Building the index takes time and memory. In return, each search is many times faster.
The question is checked against four group centres first. Then it is checked only against the points in the closest group.

Common mistake

Forgetting that the index must be rebuilt or tuned as the data grows. Results can get worse without a sound.

Key takeaways

  1. 1
    Skips most vectors. Searches where it counts.
  2. 2
    Groups or graphs. Two common kinds.
  3. 3
    Costs time to build. Saves time on search.