A Systematic Study of Semantic ID Spaces for Generative Information Retrieval
arXiv cs.CL · 7 Oct 2026, 03:33 AEST
Generative Information Retrieval (GIR) is transforming document retrieval by predicting document identifiers directly. This study presents a comprehensive analysis of effective numerical DocIDs and their properties.
Key facts
- Generative Information Retrieval (GIR) has emerged as a transformative paradigm, shifting document retrieval from a traditional 'retrieve-and-rank' workflow to sequence-to-sequence generation, where a model directly predicts document identifiers (DocIDs).
- Current approaches rely heavily on computationally expensive downstream evaluations, hindering systematic analysis and rapid iteration.
- We propose a unified framework that unifies Product Quantization (PQ) and Residual Quantization (RQ), and their hybrid variants within a single design space.
- We define a suite of training-free, intrinsic metrics, to quantify DocID quality and evaluate structural fidelity without the overhead of full model training.
- Through extensive experiments on MS MARCO 300K and NQ320K, we analyze how these structural properties influence retrieval effectiveness.
Read aloud
Generative Information Retrieval GIR is transforming document retrieval by predicting document identifiers directly. This study presents a comprehensive analysis of effective numerical DocIDs and their properties. Generative Information Retrieval GIR has emerged as a transformative paradigm, shifting document retrieval from a traditional 'retrieve-and-rank' workflow to sequence-to-sequence generation, where a model directly predicts document identifiers DocIDs. Current approaches rely heavily on computationally expensive downstream evaluations, hindering systematic analysis and rapid iteration. We propose a unified framework that unifies Product Quantization PQ and Residual Quantization RQ, and their hybrid variants within a single design space. We define a suite of training-free, intrinsic metrics, to quantify DocID quality and evaluate structural fidelity without the overhead of full model training. Through extensive experiments on MS MARCO 300K and NQ320K, we analyze how these structural properties influence retrieval effectiveness.
Sources in this cluster
- A Systematic Study of Semantic ID Spaces for Generative Information Retrieval
arXiv cs.CL · 7 Oct 2026, 03:33 AEST