ReTreever: Tree-based Coarse-to-Fine Representations for Retrieval

2025-02-11

AI Summary

Document retrieval faces challenges including high computational cost, excessive memory consumption, and poor interpretability due to high-dimensional embeddings. To address these, we propose an end-to-end trainable binary tree retrieval architecture that jointly optimizes hierarchical node routing functions to achieve semantic alignment between queries and documents along tree paths. The model implicitly learns multi-granularity semantic groupings, enabling both coarse-grained fast filtering and fine-grained precise matching. By employing path-based similarity measurement, our approach balances retrieval accuracy and efficiency: it preserves full representational capacity while outperforming existing hierarchical methods in accuracy and achieving the lowest latency. Moreover, the explicit tree structure significantly enhances system transparency and decision traceability. Overall, this work establishes a novel paradigm for efficient and interpretable document retrieval.

Technology Category

• Computer Vision: Image and Video Retrieval
• Search and Optimization: Distributed Search
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Document retrieval is a core component of question-answering systems, as it enables conditioning answer generation on new and large-scale corpora. While effective, the standard practice of encoding documents into high-dimensional embeddings for similarity search entails large memory and compute footprints, and also makes it hard to inspect the inner workings of the system. In this paper, we propose a tree-based method for organizing and representing reference documents at various granular levels, which offers the flexibility to balance cost and utility, and eases the inspection of the corpus content and retrieval operations. Our method, called ReTreever, jointly learns a routing function per internal node of a binary tree such that query and reference documents are assigned to similar tree branches, hence directly optimizing for retrieval performance. Our evaluations show that ReTreever generally preserves full representation accuracy. Its hierarchical structure further provides strong coarse representations and enhances transparency by indirectly learning meaningful semantic groupings. Among hierarchical retrieval methods, ReTreever achieves the best retrieval accuracy at the lowest latency, proving that this family of techniques can be viable in practical applications.

Problem

• Efficient document retrieval
• Memory and compute optimization
• Enhanced system transparency

Innovation

• Tree-based document representation
• Hierarchical semantic grouping
• Low-latency retrieval accuracy