MA-DPR: Manifold-aware Distance Metrics For Dense Passage Retrieval

Abstract

Dense Passage Retrieval (DPR) typically relies on Euclidean or cosine distance to measure query-passage relevance in embedding space, which is effective when embeddings lie on a linear manifold. However, our experiments across DPR benchmarks suggest that embeddings often lie on lower-dimensional, non-linear manifolds, especially in out-of-distribution (OOD) settings, where cosine and Euclidean distance fail to capture semantic similarity. To address this limitation, we propose a manifold-aware distance metric for DPR (MA-DPR) that models the intrinsic manifold structure of passages using a nearest neighbor graph and measures query-passage distance based on their shortest path in this graph. We show that MA-DPR outperforms Euclidean and cosine distances by up to 26% on OOD passage retrieval with comparable in-distribution performance across various embedding models while incurring a minimal increase in query inference time. Empirical evidence suggests that manifold-aware distance allows

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keyliu2025ma

Related papers

• Improving Dense Passage Retrieval With Multiple Positive Passages (2025)0.00
• DAPR: A Benchmark On Document-aware Passage Retrieval (2023)5.18
• Dense Passage Retrieval: Is It Retrieving? (2024)6.34
• PARM: A Paragraph Aggregation Retrieval Model For Dense Document-to-document Retrieval (2022)8.35
• Decoding Dense Embeddings: Sparse Autoencoders For Interpreting And Discretizing Dense Retrieval (2025)0.00
• Cohort Retrieval Using Dense Passage Retrieval (2025)0.00
• LIDER: An Efficient High-dimensional Learned Index For Large-scale Dense Passage Retrieval (2022)0.00
• Piecewise-linear Manifolds For Deep Metric Learning (2024)0.00