Searcharxiv⌕ Search

arXiv · 2611.19441

GRASP: Query-Adaptive Graph Retrieval framework with Cross-Attention Scoring and Path Sufficiency Verification for Question Answering

Abstract

Retrieval quality is a key factor in the reliability of retrieval-augmented generation (RAG), particularly when retrieval is performed over structured knowledge graphs. Graph-based retrieval requires decisions about candidate-edge scoring, traversal termination, and the sufficiency of the retrieved evidence. We propose GRASP, a query-adaptive graph retrieval framework that jointly addresses these decisions during the retrieval process. A cross-attention scorer models fine-grained interactions between individual query tokens and candidate edges, providing query-specific signals for edge selection. A learned stopping gate conditions traversal termination on the query, traversal state, and hop depth, allowing the retrieval depth to adapt to the current retrieval state. A sufficiency classifier further evaluates the assembled path before generation and triggers re-retrieval when the evidence is predicted to be incomplete. We evaluate GRASP on three biomedical question-answering benchmarks with different graph scales and answer lengths, using biomedical question answering as a representative setting for evaluating graph-based RAG retrieval. GRASP consistently improves over the strongest learned baseline across the evaluated retrieval and generation metrics. The unsupported-claim rate is reduced by 8.9% to 13.2% across the three benchmarks, with the largest reduction on the benchmark with the longest generated answers. The retrieval gains are more pronounced under higher candidate density, while ablation results indicate that the three components provide complementary improvements.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Rui Xu, Le Wang, Tianyue Wang. 2026-11-14. GRASP: Query-Adaptive Graph Retrieval framework with Cross-Attention Scoring and Path Sufficiency Verification for Question Answering. https://searcharxiv.com/abs/2611.19441

Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite Searcharxiv by Indiciae: BibTeX · RIS.

KEEP EXPLORING

Related papers

Practical High-Fidelity Novel-View Synthesis of Mounted Lepidoptera

Mounted butterflies are among the most striking objects in natural history collections. However, their beauty is notoriously hard to digitize in 3D: they are small and fragile, with microscopic hairs and vein structures. Capturing them in sufficient detail, therefore, requires a macro lens, which has a very limited Depth of Field (DoF). Moreover, a camera body cannot be maneuvered beneath a pinned specimen to photograph its ventral surface. We introduce an end-to-end pipeline that resolves these challenges, turning such specimens into photorealistic 3D models viewable from every direction. It combines three ingredients: handheld focus stacking for all-in-focus macro capture without a tripod, a non-contact first-surface mirror system that exposes the ventral surface without touching the specimen, and a segmentation-free, mirror-aware 3D Gaussian Splatting extension. We validate the reconstructions and design decisions on nine diverse specimens.

cs.GR↗

SSA-3DGS: Unsupervised Removal of Screen-Space Artifacts for 3D Gaussian Splatting

Novel View Synthesis (NVS) methods, such as 3D Gaussian Splatting (3DGS), rely on the assumption of clean, multi-view consistent, posed input images. Real-world captures can violate this assumption due to \textbf{screen-space artifacts}---static occlusions fixed to the 2D image plane rather than to the 3D world. Common examples include physical sensor defects, environmental obstructions (such as rain or mud on the lens enclosure), capture obstructions (such as a thumb over the camera sensor or a dashboard visible in dashcam footage), and digital overlays (such as watermarks or UI elements). When present, they are erroneously baked into the 3D geometry as ``floaters'' or near-camera artifacts, degrading the quality of novel-view rendering. In this work, we propose \textit{SSA-3DGS}, an unsupervised framework that jointly optimizes a 3D scene and a learnable 2D overlay to recover a clean 3D scene and the corrupting artifacts. By exploiting geometric consensus across views, our method effectively disentangles static artifacts from the 3D scene geometry without supervision or manual input. Across diverse synthetic corruptions and a self-captured real-world dataset, SSA-3DGS improves reconstruction fidelity by up to ${\sim}8$~dB PSNR over 3DGS trained on the same corrupted inputs, while faithfully preserving the corrupting artifact.

cs.GR↗

DOOM-GS: Dynamic Out-of-core Optimization for Massive Gaussian Splatting

Traditional Gaussian Splatting (GS) methods use in-core training, i.e. assuming that GPU memory is large enough to accommodate all the primitives. This assumption breaks down when the primitives exceed the capacity of GPU memory, e.g. GS on large environments. Recent out-of-core training research starts to address this issue, but is limited to static scenes. Unlike static GS, dynamic GS primitives move in space and time, breaking common assumptions held in existing out-of-core research and impose new challenges in memory management and training. To this end, we propose Dynamic Out-of-core Optimization for Massive Gaussian Splatting (DOOM-GS), the first out-of-core optimization method designed specifically for massive-scale dynamic GS. DOOM-GS is equipped with novel memory management strategies that optimize caching for spatio-temporal primitives, and a fast learning method for large numbers of primitives. Due to the lack of suitable datasets for large-scale dynamic GS, we also contribute a new dataset for evaluation. Since there are no directly comparable methods, we adapt several closely related methods for comparison. Through extensive evaluation, we demonstrate the effectiveness of DOOM-GS which now makes it possible to reconstruct city-scale dynamic scenes with high quality on moderate hardware, serving as a foundational tool for diverse future research and applications.

cs.GR↗