A Simple and Effective Framework for Symmetric Consistent Indexing in Large-Scale Dense Retrieval

TL;DR

This paper introduces SCI, a lightweight framework that improves large-scale dense retrieval by aligning representations and ensuring indexing consistency, thereby enhancing accuracy and stability especially for long-tail queries.

Contribution

It presents a novel symmetric representation alignment and dual-view indexing strategy that unify dual-tower representations without extra parameters, improving retrieval performance.

Findings

Improves retrieval accuracy and stability on public and e-commerce datasets.

Reduces representational misalignment and index inconsistency issues.

Supports billion-scale deployment with minimal overhead.

Abstract

Dense retrieval has become the industry standard in large-scale information retrieval systems due to its high efficiency and competitive accuracy. Its core relies on a coarse-to-fine hierarchical architecture that enables rapid candidate selection and precise semantic matching, achieving millisecond-level response over billion-scale corpora. This capability makes it essential not only in traditional search and recommendation scenarios but also in the emerging paradigm of generative recommendation driven by large language models, where semantic IDs-themselves a form of coarse-to-fine representation-play a foundational role. However, the widely adopted dual-tower encoding architecture introduces inherent challenges, primarily representational space misalignment and retrieval index inconsistency, which degrade matching accuracy, retrieval stability, and performance on long-tail queries. These issues are further magnified in semantic ID generation, ultimately limiting the performance ceiling of downstream generative models. To address these challenges, this paper proposes a simple and effective framework named SCI comprising two synergistic modules: a symmetric representation alignment module that employs an innovative input-swapping mechanism to unify the dual-tower representation space without adding parameters, and an consistent indexing with dual-tower synergy module that redesigns retrieval paths using a dual-view indexing strategy to maintain consistency from training to inference. The framework is systematic, lightweight, and engineering-friendly, requiring minimal overhead while fully supporting billion-scale deployment. We provide theoretical guarantees for our approach, with its effectiveness validated by results across public datasets and real-world e-commerce datasets.