Elastic Index Selection for Label-Hybrid AKNN Search

TL;DR

This paper proposes an elastic index selection method for label-hybrid approximate k-nearest neighbor search, leveraging label set inclusion to improve efficiency and scalability while maintaining performance.

Contribution

It introduces elastic factor bounds and a greedy algorithm for index selection, enabling efficient, scalable label-hybrid AKNN search with performance guarantees.

Findings

Achieves up to 500x search speedup over state-of-the-art methods.

Demonstrates effectiveness on multiple real datasets.

Provides a versatile, library-compatible solution.

Abstract

Real-world vector embeddings are usually associated with extra labels, such as attributes and keywords. Many applications require the nearest neighbor search that contains specific labels, such as searching for product image embeddings restricted to a particular brand. A straightforward approach is to materialize all possible indices according to the complete query label workload. However, this leads to an exponential increase in both index space and processing time, which significantly limits scalability and efficiency. In this paper, we leverage the inclusion relationships among query label sets to construct partial indexes, enabling index sharing across queries for improved construction efficiency. We introduce \textit{elastic factor} bounds to guarantee search performance and use the greedy algorithm to select indices that meet the bounds, achieving a tradeoff between efficiency and space. Meanwhile, we also designed the algorithm to achieve the best elastic factor under a given space limitation. Experimental results on multiple real datasets demonstrate that our algorithm can achieve near-optimal search performance, achieving up to 10x-500x search efficiency speed up over state-of-the-art approaches. Our algorithm is highly versatile, since it is not constrained by index type and can seamlessly integrate with existing optimized libraries.