Personalized kNN Query Execution in Vector Databases

D 2026

Personalized kNN Query Execution in Vector Databases

ŠIKYŇA, Matúš a Pavel ZEZULA

Základní údaje

Originální název

Personalized kNN Query Execution in Vector Databases

Autoři

ŠIKYŇA, Matúš a Pavel ZEZULA

Vydání

New York, NY, USA, Proceedings of the 35th ACM International Conference on Information and Knowledge Management (CIKM '26), 11 s. 2026

Nakladatel

Association for Computing Machinery

Další údaje

Jazyk

angličtina

Typ výsledku

Stať ve sborníku

Obor

10200 1.2 Computer and information sciences

Stát vydavatele

Spojené státy

Utajení

není předmětem státního či obchodního tajemství

Forma vydání

elektronická verze "online"

Označené pro přenos do RIV

Organizační jednotka

Fakulta informatiky

Klíčová slova anglicky

Personalized Search; Metric Learning; Vector Databases

Štítky

DISA, similarity search

Příznaky

Mezinárodní význam, Recenzováno

Změněno: 15. 9. 2026 06:50, RNDr. Mgr. Matúš Šikyňa

Anotace

V originále

Modern vector databases enable search over high-dimensional embeddings, but query execution engines rely on a specific similarity metric that cannot reflect how individual users perceive similarity. We advocate a metric-level perspective on personalization, keeping the shared embedding space and the underlying index unchanged, and representing each user's specific search needs by a small matrix for Mahalanobis distance learned using relevance feedback and metric learning. Using a lower bounding relationship between Mahalanobis and Euclidean distances, we propose a query execution strategy for personalized k-nearest neighbour search that retrieves an enlarged candidate set using a standard Euclidean (or inner product) index and refines it under the user’s subjective metric. Experiments on CLIP embeddings from Profiset and LAION-2B-en datasets using FAISS-based indices show that this approach can be layered on top of existing vector databases with predictable behaviour, high overlap with precise Mahalanobis kNN search results, and only a small overhead over standard Euclidean search. When the Euclidean–Mahalanobis discrepancy becomes too large for efficient refinement, we propose approximation techniques and show that a Cholesky-based whitening transformation can reorganize embeddings so that personalized search reduces to Euclidean indexing again.

Návaznosti

MUNI/A/1860/2025, interní kód MU

Název: Zapojení studentů Fakulty informatiky do mezinárodní vědecké komunity 26

Investor: Masarykova univerzita, Zapojení studentů Fakulty informatiky do mezinárodní vědecké komunity 26

MUNI/A/1873/2025, interní kód MU

Název: Komplexní data a systémy s využitím umělé inteligence

Investor: Masarykova univerzita, Komplexní data a systémy s využitím umělé inteligence

VK01010147, projekt VaV

Název: Automatizovaná forenzní laboratoř digitálních dat pro odhalování komplexní trestné činnosti

Investor: Ministerstvo vnitra ČR, Automatizovaná forenzní laboratoř digitálních dat pro odhalování komplexní trestné činnosti

Citovat

ŠIKYŇA, Matúš a Pavel ZEZULA. Personalized kNN Query Execution in Vector Databases. Online. In Proceedings of the 35th ACM International Conference on Information and Knowledge Management (CIKM '26). New York, NY, USA: Association for Computing Machinery, 2026, 11 s.

@inproceedings{2588718,
   author = {Šikyňa, Matúš and Zezula, Pavel},
   address = {New York, NY, USA},
   booktitle = {Proceedings of the 35th ACM International Conference on Information and Knowledge Management (CIKM '26)},
   keywords = {Personalized Search; Metric Learning; Vector Databases},
   howpublished = {elektronická verze "online"},
   language = {eng},
   location = {New York, NY, USA},
   publisher = {Association for Computing Machinery},
   title = {Personalized kNN Query Execution in Vector Databases},
   year = {2026}
}

TY  - CONF
ID  - 2588718
AU  - Šikyňa, Matúš - Zezula, Pavel
PY  - 2026
TI  - Personalized kNN Query Execution in Vector Databases
PB  - Association for Computing Machinery
CY  - New York, NY, USA
KW  - Personalized Search
KW  - Metric Learning
KW  - Vector Databases
N2  - Modern vector databases enable search over high-dimensional embeddings, but query execution engines rely on a specific similarity metric that cannot reflect how individual users perceive similarity. We advocate a metric-level perspective on personalization, keeping the shared embedding space and the underlying index unchanged, and representing each user's specific search needs by a small matrix for Mahalanobis distance learned using relevance feedback and metric learning. Using a lower bounding relationship between Mahalanobis and Euclidean distances, we propose a query execution strategy for personalized k-nearest neighbour search that retrieves an enlarged candidate set using a standard Euclidean (or inner product) index and refines it under the user’s subjective metric. Experiments on CLIP embeddings from Profiset and LAION-2B-en datasets using FAISS-based indices show that this approach can be layered on top of existing vector databases with predictable behaviour, high overlap with precise Mahalanobis kNN search results, and only a small overhead over standard Euclidean search. When the Euclidean–Mahalanobis discrepancy becomes too large for efficient refinement, we propose approximation techniques and show that a Cholesky-based whitening transformation can reorganize embeddings so that personalized search reduces to Euclidean indexing again.
ER  -

ŠIKYŇA, Matúš a Pavel ZEZULA. Personalized kNN Query Execution in Vector Databases. Online. In \textit{Proceedings of the 35th ACM International Conference on Information and Knowledge Management (CIKM '26)}. New York, NY, USA: Association for Computing Machinery, 2026, 11 s.

Přehled o publikaci