Sparse, Dense, and Attentional Representations for Text Retrieval

TL;DR

This paper analyzes the capacity of dual encoders for text retrieval, revealing limitations with long documents, and proposes hybrid models that combine sparse and dense representations to improve large-scale retrieval performance.

Contribution

It provides theoretical and empirical insights into the limitations of fixed-length encodings and introduces hybrid models that enhance retrieval accuracy and efficiency.

Findings

Hybrid models outperform strong baselines in large-scale retrieval.

Limitations of fixed-length encodings increase with document length.

Combining sparse and dense representations improves retrieval precision.

Abstract

Dual encoders perform retrieval by encoding documents and queries into dense lowdimensional vectors, scoring each document by its inner product with the query. We investigate the capacity of this architecture relative to sparse bag-of-words models and attentional neural networks. Using both theoretical and empirical analysis, we establish connections between the encoding dimension, the margin between gold and lower-ranked documents, and the document length, suggesting limitations in the capacity of fixed-length encodings to support precise retrieval of long documents. Building on these insights, we propose a simple neural model that combines the efficiency of dual encoders with some of the expressiveness of more costly attentional architectures, and explore sparse-dense hybrids to capitalize on the precision of sparse retrieval. These models outperform strong alternatives in large-scale retrieval.