Wikitopia
HNSW
conceptAlgorithm
Try in Playground →RSS
Overview
Developed byYu. A. Malkov
Maintained byNMSLIB team
Founded2016
LicenseApache License 2.0
Open source✓ Open Source
Use casesimilarity search in high-dimensional spaces
Knowledge graph stats
Claims70
Avg confidence92%
Avg freshness99%
Last updatedUpdated 5 days ago
WikidataQ124785393
Trust distribution
100% unverified
Governance
Not assessed
Contribute governance data →

HNSW

concept

Hierarchical Navigable Small World algorithm for approximate nearest neighbor search in high dimensions.

Compare with...

primary use case

ValueTrustConfidenceFreshnessSources
similarity search in high-dimensional spaces○UnverifiedHighFresh1
approximate nearest neighbor search○UnverifiedHighFresh1
approximate nearest neighbor search in high-dimensional spaces○UnverifiedHighFresh1
high-dimensional vector similarity search○UnverifiedHighFresh1
vector similarity search○UnverifiedHighFresh1
machine learning embedding search○UnverifiedHighFresh1

supports model

ValueTrustConfidenceFreshnessSources
embedding vectors○UnverifiedHighFresh1
vector embeddings○UnverifiedHighFresh1

integrates with

ValueTrustConfidenceFreshnessSources
Hnswlib○UnverifiedHighFresh1
C++○UnverifiedHighFresh1
Python○UnverifiedHighFresh1
Weaviate○UnverifiedHighFresh1
Chroma○UnverifiedHighFresh1
Faiss○UnverifiedModerateFresh1

based on

ValueTrustConfidenceFreshnessSources
navigable small world graphs○UnverifiedHighFresh1
hierarchical navigable small world graphs○UnverifiedHighFresh1
navigable small world networks○UnverifiedHighFresh1
skip list data structure○UnverifiedHighFresh1
small world networks○UnverifiedHighFresh1

supports data type

ValueTrustConfidenceFreshnessSources
high-dimensional vectors○UnverifiedHighFresh1

algorithm type

ValueTrustConfidenceFreshnessSources
graph-based search algorithm○UnverifiedHighFresh1

supports protocol

ValueTrustConfidenceFreshnessSources
vector similarity search○UnverifiedHighFresh1
euclidean distance○UnverifiedHighFresh1
cosine similarity○UnverifiedHighFresh1
high-dimensional vector spaces○UnverifiedHighFresh1

alternative to

ValueTrustConfidenceFreshnessSources
brute force k-NN search○UnverifiedHighFresh1
LSH○UnverifiedHighFresh1
KD-tree○UnverifiedModerateFresh1
LSH (Locality-Sensitive Hashing)○UnverifiedModerateFresh1
IVF (Inverted File Index)○UnverifiedModerateFresh1
k-d trees○UnverifiedModerateFresh1
Annoy○UnverifiedModerateFresh1
LSH (Locality Sensitive Hashing)○UnverifiedModerateFresh1
k-d tree○UnverifiedModerateFresh1

used in domain

ValueTrustConfidenceFreshnessSources
vector similarity search○UnverifiedHighFresh1
vector databases○UnverifiedHighFresh1
machine learning○UnverifiedHighFresh1

data structure type

ValueTrustConfidenceFreshnessSources
graph-based index○UnverifiedHighFresh1

developed by

ValueTrustConfidenceFreshnessSources
Yu. A. Malkov○UnverifiedHighFresh1
D. A. Yashunin○UnverifiedHighFresh1
Yury Malkov○UnverifiedHighFresh1
Yu. A. Malkov and D. A. Yashunin○UnverifiedHighFresh1

founded year

ValueTrustConfidenceFreshnessSources
2016○UnverifiedHighFresh1

published year

ValueTrustConfidenceFreshnessSources
2016○UnverifiedHighFresh1

supports use case

ValueTrustConfidenceFreshnessSources
vector similarity search○UnverifiedHighFresh1
recommendation systems○UnverifiedModerateFresh1

time complexity

ValueTrustConfidenceFreshnessSources
O(log N) expected○UnverifiedHighFresh1

implemented in

ValueTrustConfidenceFreshnessSources
C++○UnverifiedHighFresh1

supports distance metric

ValueTrustConfidenceFreshnessSources
cosine similarity○UnverifiedHighFresh1
Euclidean distance○UnverifiedHighFresh1

used in

ValueTrustConfidenceFreshnessSources
machine learning applications○UnverifiedHighFresh1

use case

ValueTrustConfidenceFreshnessSources
machine learning embeddings search○UnverifiedHighFresh1

supports metric

ValueTrustConfidenceFreshnessSources
Euclidean distance○UnverifiedHighFresh1
cosine similarity○UnverifiedHighFresh1

license type

ValueTrustConfidenceFreshnessSources
Apache License 2.0○UnverifiedHighFresh1
Apache 2.0○UnverifiedModerateFresh1

requires

ValueTrustConfidenceFreshnessSources
metric space○UnverifiedHighFresh1

open source

ValueTrustConfidenceFreshnessSources
true○UnverifiedHighFresh1

algorithm complexity

ValueTrustConfidenceFreshnessSources
O(log N) search time○UnverifiedHighFresh1

supports language

ValueTrustConfidenceFreshnessSources
C++○UnverifiedHighFresh1
Python○UnverifiedHighFresh1

maintained by

ValueTrustConfidenceFreshnessSources
NMSLIB team○UnverifiedModerateFresh1
nmslib community○UnverifiedModerateFresh1

competes with

ValueTrustConfidenceFreshnessSources
FAISS○UnverifiedModerateFresh1
IVF (Inverted File)○UnverifiedModerateFresh1
ScaNN○UnverifiedModerateFresh1
IVF (Inverted File Index)○UnverifiedModerateFresh1
Annoy○UnverifiedModerateFresh1

Alternatives & Similar Tools

brute force k-NN search
alternative to
Compare →
LSH
alternative to
Compare →
KD-tree
alternative to
Compare →
LSH (Locality-Sensitive Hashing)
alternative to
Compare →
FAISS
competes with
Compare →
IVF (Inverted File Index)
alternative to
Compare →
IVF (Inverted File)
competes with
Compare →
k-d trees
alternative to
Compare →
ScaNN
competes with
Compare →
Annoy
alternative to
Compare →
IVF (Inverted File Index)
competes with
Compare →
Annoy
competes with
Compare →
LSH (Locality Sensitive Hashing)
alternative to
Compare →
k-d tree
alternative to
Compare →

Commonly Used With

HnswlibC++PythonWeaviateChromaFaiss

Related entities

Claim count: 70Last updated: 4/5/2026Edit history