AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

Improving Long-Context Summarization with Multi-Granularity Retrieval Optimization

Xueyu Chen, Kaitao Song, Zifan Song, Dongsheng Li, Cairong Zhao

AAAI 2026 · Large language models › Retrieval and knowledge · approach: RAG robustness and evaluation

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add HTSIR summary tree on top of DPR retrieval (ablation)DPR retriever + GPT-4o mini · QuALITY-HARDaccuracy: 55.31 -> 62.57Improved
add HTSIR summary tree on top of BM25 retrieval (ablation)BM25 retriever + GPT-4o mini · NarrativeQAROUGE-L: 26.52 -> 30.47Improved
add HTSIR summary tree on top of SBERT retrieval (ablation)SBERT retriever + GPT-4o mini · QASPERAnswer F1: 32.65 -> 34.48Improved
retrieve all three node layers (leaf, intermediate, root) instead of one layer (ablation)HTSIR tree · QuALITY-HARDaccuracy: 61.54 -> 62.50Improved
retrieve two node layers instead of one (ablation)HTSIR tree · QuALITY-HARDaccuracy: 60.57 -> 57.69 (x vs x+s)Hurt
retrieve only root summary nodes (ablation)HTSIR tree · QuALITY-HARDaccuracy: 62.50 -> 54.81Hurt
vary r (chunks per summary) from 2 to 5 (ablation)HTSIR · QuALITY-HARDaccuracy: 62.48 -> 62.99 (r=2 to r=4)No clear effect
prepend HTSIR retrieved content to full text for GPT-4o miniGPT-4o mini · QuALITY-HARDaccuracy: 72.1 -> 74.0Improved
use HTSIR with Llama3-8B LoRA instead of plain Llama3-8B LoRALlama3-8B LoRA · QASPERAnswer F1: 21.7 -> 23.5Improved
use HTSIR tree instead of RAPTOR clustering treetree-based RAG · QASPERAnswer F1: 36.7 -> 36.9No clear effect
use HTSIR instead of graph-based RAG (HippoRAG 2)long-document QA · NarrativeQAF1: 25.2 -> 37.4Improved
use fixed-length chunk summaries without a tree, at tuned length (150 x 5 or 100 x 7 chunks) (ablation)summary corpus without tree · QASPERAnswer F1Mixed

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy