Billion-scale Similarity Search Using a Hybrid Indexing Approach with Advanced Filtering
This work addresses the need for efficient large-scale similarity search with filtering capabilities, primarily for CPU-based systems, but it appears incremental as it builds on existing IVF-Flat methods.
The paper tackles the problem of similarity search with complex filtering on billion-scale datasets, presenting a hybrid indexing approach that extends IVF-Flat to integrate multi-dimensional filters, resulting in a cost-effective CPU-optimized solution.
This paper presents a novel approach for similarity search with complex filtering capabilities on billion-scale datasets, optimized for CPU inference. Our method extends the classical IVF-Flat index structure to integrate multi-dimensional filters. The proposed algorithm combines dense embeddings with discrete filtering attributes, enabling fast retrieval in high-dimensional spaces. Designed specifically for CPU-based systems, our disk-based approach offers a cost-effective solution for large-scale similarity search. We demonstrate the effectiveness of our method through a case study, showcasing its potential for various practical uses.