Forward Index Compression for Learned Sparse Retrieval

TL;DR

The paper addresses the forward index size in learned sparse retrieval, where the forward index $\mathcal{X}$ stores sparse document vectors and is used to compute the inner product $\langle q, x \rangle$ with queries. To reduce memory while preserving query latency, it evaluates a suite of integer compression methods for the forward-index components and identifies StreamVByte as the strong baseline. The main contribution is DotVByte, a SIMD-optimized variant tailored to inner-product computation that achieves significant space savings with minimal impact on retrieval quality; experiments on MsMarco with Splade and LiLsr show substantial improvements. Code is released in the Seismic library as Rust implementations, enabling practical deployment.

Abstract

Text retrieval using learned sparse representations of queries and documents has, over the years, evolved into a highly effective approach to search. It is thanks to recent advances in approximate nearest neighbor search-with the emergence of highly efficient algorithms such as the inverted index-based Seismic and the graph-based Hnsw-that retrieval with sparse representations became viable in practice. In this work, we scrutinize the efficiency of sparse retrieval algorithms and focus particularly on the size of a data structure that is common to all algorithmic flavors and that constitutes a substantial fraction of the overall index size: the forward index. In particular, we seek compression techniques to reduce the storage footprint of the forward index without compromising search quality or inner product computation latency. In our examination with various integer compression techniques, we report that StreamVByte achieves the best trade-off between memory footprint, retrieval accuracy, and latency. We then improve StreamVByte by introducing DotVByte, a new algorithm tailored to inner product computation. Experiments on MsMarco show that our improvements lead to significant space savings while maintaining retrieval efficiency.