内容库 / 论文

arXiv:2608.21252 · RAG 检索增强

EnSI-RAG: Entity-Structure-Indexed Retrieval-Augmented Generation for Long-Document Question Answering

EnSI-RAG：面向长文档问答的 Entity-Structure-Indexed RAG

EnSI-RAG: Entity-Structure-Indexed Retrieval-Augmented Generation for Long-Document Question Answering

类型：arxiv
标识：2608.21252
链接：http://arxiv.org/abs/2608.21252v1
主分类：rag
形态：method
TLDR：Question answering (QA) over long, connected documents remains challenging because relevant evidence may span multiple entities and their relationships. Existing retrieval-augmented generation (RAG) methods typically index documents as raw chunks and retrieve them through embedding similarity. Their performance degrades when chunk boundaries separate entities from supporting evidence or when a question requires multi-hop reasoning across the corpus. We propose EnSI-RAG (Entity-Structure-Indexed Retrieval-Augmented Generation), a framework that constructs a query-independent, entity-centered in
待LLM分类：否
标题中文：EnSI-RAG：面向长文档问答的 Entity-Structure-Indexed RAG
TLDR中文：长篇连通文档上的 QA 仍具挑战，因为相关证据可能跨越多个实体及其关系。现有 RAG 方法通常将文档以原始 chunk 索引并通过 embedding 相似度检索，当 chunk 边界切断实体与支持证据的联系，或问题需在语料库中多跳推理时性能下降。我们提出 EnSI-RAG（Entity-Structure-Indexed RAG），一个构建 query-independent、以实体为中心的[索引]框架……
来源文件：
/inbox/tom/_candidates/2026-08-25-agent-rag-longcontext-candidates.json

相关 · 按语义匹配

笔记 Cool Papers · cs.IR (papers.cool) · RSS 摘要

笔记 Cool Papers · cs.CL (papers.cool) · RSS 摘要

笔记 Tom 文献雷达 · AI Agent / RAG / 长上下文 · 2026-08-03 晚间

笔记精读与批判 · Prompt-Model Interaction Reaches the Fixed Points (arXiv:2608.21315)

笔记 flyP 精读草稿 · 2026-06-18 · 多模态位置证据与长上下文检索

仓库 oigrex666/mcp-context-chronicler

arXiv

2608.21252

TLDR

长篇连通文档上的 QA 仍具挑战，因为相关证据可能跨越多个实体及其关系。现有 RAG 方法通常将文档以原始 chunk 索引并通过 embedding 相似度检索，当 chunk 边界切断实体与支持证据的联系，或问题需在语料库中多跳推理时性能下降。我们提出 EnSI-RAG（Entity-Structure-Indexed RAG），一个构建 query-independent、以实体为中心的[索引]框架……Question answering (QA) over long, connected documents remains challenging because relevant evidence may span multiple entities and their relationships. Existing retrieval-augmented generation (RAG) methods typically index documents as raw chunks and retrieve them through embedding similarity. Their performance degrades when chunk boundaries separate entities from supporting evidence or when a question requires multi-hop reasoning across the corpus. We propose EnSI-RAG (Entity-Structure-Indexed Retrieval-Augmented Generation), a framework that constructs a query-independent, entity-centered in

相关解读

arXiv ↗ 中文全文