Skip to main content
Loading

Vector glossary

It is important to become familiar with the following terms before exploring the Aerospike Vector Search (AVS) documentation.

Vector terminology​

The following vector-related terms are important to understanding AVS concepts.

ANN​

Approximate Nearest Neighbor search is a classification of search that attempts to get the best approximate results in a short amount of time.

Dimensions (dims)​

The number of elements that make up a vector. These elements can be floats, integers, or binary digits.

HNSW​

Hierarchical Navigable Small World (HNSW) is an algorithm for efficient approximate nearest neighbor search in high-dimensional spaces.

Index​

A data structure or mechanism that improves the speed and efficiency of data retrieval operations within a database or other data storage systems.

Index cache​

A cached version of index records held in AVS memory. For more details, see search caching.

Index metadata (Namespace)​

Data captured about an index. Using AVS requires configuring an index metadata namespace in the Aerospike Database storage layer.

Index record​

An individual index record traverses the HNSW index. For more details, see index records

(exact) KNN​

An exact k Nearest Neighbor search, which is an exhaustive search technique that gets the best results but can take a long time to perform.

RAG​

Retrieval-augmented Generation (RAG) is a model that combines retrieval-based and generation-based approaches to generate more accurate and informative responses by augmenting generative models with retrieved documents.

Storage layer​

The Aerospike Database storage system used by AVS.

Unmerged records​

Records that have not been added to an index, which can be monitored using asvec.

Vector embedding​

A numerical representation of data in a high-dimensional space, capturing essential features and relationships.

Vector metadata​

User-supplied data that is written in records along with vector embeddings.

Aerospike terminology​

The following are Aerospike-specific terms and their relational equivalents:

TermRelational EquivalentDescription
NamespaceDatabaseA logical separation of data within Aerospike. A namespace allows an administrator to define storage, availability, and consistency properties for the data within it.
SetTableAn organizational separation of data within Aerospike. A set allows an administrator to monitor usage of data within it.
(Vector) bin/fieldColumnThe field from which you want to create your index.
(Vector) recordRowAll the data associated with a vector. This includes the vector itself, any metadata associated with the record, and a primary key for retrieving the record.