Sparse, Dense, and Attentional Representations for Text Retrieval

• Published in: Transactions of the Association for Computational Linguistics Vol. 9; pp. 329 - 345
• Main Authors: Luan, Yi, Eisenstein, Jacob, Toutanova, Kristina, Collins, Michael
• Format: Journal Article
• Language: English
• Published: One Rogers Street, Cambridge, MA 02142-1209, USA MIT Press 01.01.2021; MIT Press Journals, The; The MIT Press
• Subjects: Attention; Benchmarks; Coders; Coding; Computational linguistics; Documents; Empirical analysis; Linguistics; Neural networks; Queries; Retrieval
• ISSN: 2307-387X; 2307-387X
• DOI: 10.1162/tacl_a_00369

Dual encoders perform retrieval by encoding documents and queries into dense low-dimensional 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.