aiqbench analysis
This research presents a parameterized framework for fusing dense and sparse retrieval signals, demonstrating that dataset-specific tuning of mixing coefficients yields measurable gains over standard equal-weight baselines on scientific fact datasets. The study highlights the limitations of generic heuristics like uniform RRF smoothing while providing a reproducible grid-search protocol for optimizing hybrid rankers in information retrieval tasks.
Parameterized Dense-Sparse Fusion for Hybrid Retrieval: Tuning a Rank-Score Mix on BEIR SciFact with Qdrant
arXiv:2609.22770v2 Announce Type: replace-cross Abstract: We study a parameterized hybrid ranker that fuses a dense embedding list and a sparse lexical list. The method has a small, explicit parameter vector: a dense prior $\alpha \in [0,1]$, a score-versus-rank mix $\lambda \in [0,1]$, an RRF smoothing parameter $\kappa > 0$, optional list-geometry coefficients that move $\alpha$ per query, and a router margin $\tau$ that can turn sparse search off. We grid-search those ranges on SciFact train (809 queries) and freeze the chosen values on SciFact test (300). The tuned rank-score mix ($\alpha = 0.8$, $\lambda = 0.75$, $\kappa = 20$) reaches 0.753 nDCG@10 and 0.889 recall@10, outperforming dense BGE (0.742 / 0.871) and equal-weight RRF (0.707 nDCG@10) on that test split. A list-conditioned $\alpha$ adds +0.0006 nDCG; a sparse-off router is rejected by the same train split (any $\tau$ that skipped approximately 50% of queries lost nDCG). These coefficients are dataset-specific. Equal RRF with the same models does not beat dense on a nine-zip BEIR macro-average (0.479 vs. 0.519 nDCG@10). Repeating the same train-then-freeze sweep independently on all 20 indexed units beats equal RRF on 20/20 and dense on 16/20 (unit-mean nDCG@10 0.467 vs. 0.462 dense vs. 0.420 RRF). Other corpora should reuse the ranges, not a copy of the SciFact point.
Abstract:We study a parameterized hybrid ranker that fuses a dense embedding list and a sparse lexical list. The method has a small, explicit parameter vector: a dense prior $\alpha \in [0,1]$, a score-versus-rank mix $\lambda \in [0,1]$, an RRF smoothing parameter $\kappa > 0$, optional list-geometry coefficients that move $\alpha$ per query, and a router margin $\tau$ that can turn sparse search off. We grid-search those ranges on SciFact train (809 queries) and freeze the chosen values on SciFact test (300). The tuned rank-score mix ($\alpha = 0.8$, $\lambda = 0.75$, $\kappa = 20$) reaches 0.753 nDCG@10 and 0.889 recall@10, outperforming dense BGE (0.742 / 0.871) and equal-weight RRF (0.707 nDCG@10) on that test split. A list-conditioned $\alpha$ adds +0.0006 nDCG; a sparse-off router is rejected by the same train split (any $\tau$ that skipped approximately 50% of queries lost nDCG). These coefficients are dataset-specific. Equal RRF with the same models does not beat dense on a nine-zip BEIR macro-average (0.479 vs. 0.519 nDCG@10). Repeating the same train-then-freeze sweep independently on all 20 indexed units beats equal RRF on 20/20 and dense on 16/20 (unit-mean nDCG@10 0.467 vs. 0.462 dense vs. 0.420 RRF). Other corpora should reuse the ranges, not a copy of the SciFact point.
Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
arXiv:2609.22770 [cs.IR]
(or arXiv:2609.22770v2 [cs.IR] for this version)
https://doi.org/10.48550/arXiv.2609.22770 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 301 words
Related coverage
Multimodal Hybrid Retrieval-Augmented Generation for Scientific Document Understanding using Open-Source SLMs
arXiv · Jul 29, 2026
Dense Expands, Sparse Anchors: Channel-Asymmetric Query Expansion for Hybrid Retrieval
arXiv · Aug 18, 2026
GEM Training: How Meta Doubled the Efficiency of Its LLM-Scale Ads Foundation Model
Meta Engineering · Aug 3, 2026
Semantic Candidate-Job Matching: A Comparative Evaluation of Dense Embedding Models in Hybrid Retrieval
arXiv · Sep 22, 2026