Coarse Indexing, Fine Evidence: Decoupling Temporal Granularity in Long-Video RAG

• URL: http://arxiv.org/abs/2608.23011v1
• Date: Mon, 24 Aug 2026 09:12:06 GMT
• Title: Coarse Indexing, Fine Evidence: Decoupling Temporal Granularity in Long-Video RAG
• Abstract summary: We propose textbfDensity-Aware Graph Construction (DAGC), a training-free approach that decouples a query-independent coarse retrieval index from the original fine-grained evidence space.<n> Experiments on MLVU, VideoMME, and LongVideoBench show that DAGC retains only about 40--50% of the original graph nodes and achieves $1.3$--$1.7times$ end-to-end wall-clock acceleration.<n>The gains transfer across different LVLM backbones and video RAG pipelines, suggesting that long-video RAG need not maintain the same temporal
• Score: 9.939361774115874
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Graph-based retrieval-augmented generation (RAG) provides a scalable paradigm for long-video understanding, but existing systems typically inherit a fixed temporal granularity from video segmentation when constructing their retrieval index. We argue that this design unnecessarily couples indexing granularity with evidence granularity: coarse representations can often suffice for locating relevant temporal regions, while fine-grained evidence remains important for downstream reasoning. We propose \textbf{Density-Aware Graph Construction (DAGC)}, a training-free approach that decouples a query-independent coarse retrieval index from the original fine-grained evidence space. DAGC constructs a compact, density-adaptive graph index by merging visually redundant neighboring chunks, while preserving mappings to the original temporal units. Retrieved coarse regions are subsequently expanded back to the original chunk granularity for fine-grained evidence refinement and answer generation. Experiments on MLVU, VideoMME, and LongVideoBench show that DAGC retains only about 40--50\% of the original graph nodes and achieves $1.3$--$1.7\times$ end-to-end wall-clock acceleration while preserving approximately 99\% of the original QA performance. The gains transfer across different LVLM backbones and video RAG pipelines, suggesting that long-video RAG need not maintain the same temporal granularity for indexing and evidence reasoning.

Related papers

• Efficient Retrieval-Augmented Generation via Token Co-occurrence Graphs [6.058906522832935]
  TIGRAG (Token-Induced GraphRAG) is an efficient graph-augmented RAG framework based on a token co-occurrence Knowledge Graph.<n>During inference, it combines graph-based semantic expansion and neural reranking to retrieve interconnected evidence for multi-hop reasoning.
  arXiv  Detail & Related papers  (2026-06-29T10:29:51Z)
• Rethinking RAG in Long Videos: What to Retrieve and How to Use It? [56.38819694781005]
  We introduce V-RAGBench, a benchmark of $langle$query, evidence chunk, answer$rangle$ triplets that enables faithful, decoupled evaluation of retrieval and generation.<n>We also present CARVE, a method that runs parallel retrievers across configurations and employs chunk-adaptive reranking to identify the winning configuration for each chunk.
  arXiv  Detail & Related papers  (2026-06-11T10:05:49Z)
• Beyond Chunk-Local Extraction: Cross-Chunk Graph Augmentation for GraphRAG [11.699085890492833]
  GraphRAG extends retrieval-augmented generation by organizing corpora as explicit knowledge graphs.<n>Existing frameworks extract entities and relations within individual chunks, leaving cross-chunk relations systematically absent from the index.<n>We present CrossAug, a GNN-guided CROSS-Chunk Graph AUGmentation method that enriches GraphRAG indices with cross-chunk relational structure.
  arXiv  Detail & Related papers  (2026-05-27T05:55:53Z)
• Attention Itself Could Retrieve.RetrieveVGGT: Training-Free Long Context Streaming 3D Reconstruction via Query-Key Similarity Retrieval [86.05617542701644]
  Visual Geometry Grounded Transformer (VGGT) advances 3D reconstruction via scalable Transformer architecture.<n>StreamVGGT enables streaming with causal attention, yet its KV cache grows linearly with frames, causing memory overflow and quality degradation.<n>We present RetrieveVGGT, a training-free framework, which formulates context construction for VGGT as a retrieval problem.
  arXiv  Detail & Related papers  (2026-05-10T16:41:54Z)
• Event-Causal RAG: A Retrieval-Augmented Generation Framework for Long Video Reasoning in Complex Scenarios [9.729442664774988]
  Event-Causal RAG is a lightweight retrieval-augmented framework for infinite long-video reasoning.<n>Method segments streaming videos into semantically coherent events and represents each event as a structured State-Event-State graph.<n>On top of this memory, we design a bidirectional retrieval strategy to efficiently identify the most relevant event causal chains.
  arXiv  Detail & Related papers  (2026-05-07T13:01:28Z)
• VideoAtlas: Navigating Long-Form Video in Logarithmic Compute [3.705718227493618]
  textbfVideoAtlas is a task-agnostic environment to represent video as a hierarchical grid.<n>The hierarchical structure ensures access depth grows only logarithmically with video length.<n>Video-RLM remains the most duration-robust method with minimal accuracy degradation when scaling from 1-hour to 10-hour benchmarks.
  arXiv  Detail & Related papers  (2026-03-18T17:20:19Z)
• Right Answer at the Right Time - Temporal Retrieval-Augmented Generation via Graph Summarization [23.1799368651364]
  Question answering in temporal knowledge graphs requires retrieval that is both time-consistent and efficient.<n>We propose STAR-RAG, a temporal GraphRAG framework that relies on two key ideas: building a time-aligned rule graph and conducting propagation on this graph to narrow the search space.<n>This design enforces temporal proximity during retrieval, reduces the candidate set of retrieval results, and lowers token consumption without sacrificing accuracy.
  arXiv  Detail & Related papers  (2025-10-19T05:00:04Z)
• Vgent: Graph-based Retrieval-Reasoning-Augmented Generation For Long Video Understanding [56.45689495743107]
  Vgent is a graph-based retrieval-reasoning-augmented generation framework to enhance LVLMs for long video understanding.<n>We evaluate our framework with various open-source LVLMs on three long-video understanding benchmarks.
  arXiv  Detail & Related papers  (2025-10-15T19:14:58Z)
• Respecting Temporal-Causal Consistency: Entity-Event Knowledge Graphs for Retrieval-Augmented Generation [69.45495166424642]
  We develop a robust and discriminative QA benchmark to measure temporal, causal, and character consistency understanding in narrative documents.<n>We then introduce Entity-Event RAG (E2RAG), a dual-graph framework that keeps separate entity and event subgraphs linked by a bipartite mapping.<n>Across ChronoQA, our approach outperforms state-of-the-art unstructured and KG-based RAG baselines, with notable gains on causal and character consistency queries.
  arXiv  Detail & Related papers  (2025-06-06T10:07:21Z)
• DIFFVSGG: Diffusion-Driven Online Video Scene Graph Generation [61.59996525424585]
  DIFFVSGG is an online VSGG solution that frames this task as an iterative scene graph update problem.<n>We unify the decoding of object classification, bounding box regression, and graph generation three tasks using one shared feature embedding.<n>DIFFVSGG further facilitates continuous temporal reasoning, where predictions for subsequent frames leverage results of past frames as the conditional inputs of LDMs.
  arXiv  Detail & Related papers  (2025-03-18T06:49:51Z)
• Temporally Consistent Referring Video Object Segmentation with Hybrid Memory [98.80249255577304]
  We propose an end-to-end R-VOS paradigm that explicitly models temporal consistency alongside the referring segmentation. Features of frames with automatically generated high-quality reference masks are propagated to segment remaining frames. Extensive experiments demonstrate that our approach enhances temporal consistency by a significant margin.
  arXiv  Detail & Related papers  (2024-03-28T13:32:49Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.