# MDKeyChunker: Single-Call LLM Enrichment with Rolling Keys and Key-Based Restructuring for High-Accuracy RAG

arXiv: https://arxiv.org/abs/2603.23533 · HTML: https://mlsanity.com/paper/2603.23533/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Multimodal RAG, one of 15 directions in Retrieval-augmented generation.
- Builds on 18 of its 37 references that are in the map.
- First posted on arXiv 8 Mar 2026.

## Placed in

- Retrieval-augmented generation / [Multimodal RAG](https://mlsanity.com/rag/multimodal-rag/)

## Builds on (most cited first, 18 of 18)

- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [From Local to Global: A Graph RAG Approach to Query-Focused Summarization](https://mlsanity.com/paper/2404.16130/) (2024)
- [Improving language models by retrieving from trillions of tokens](https://mlsanity.com/paper/2112.04426/) (2021)
- [RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrieval](https://mlsanity.com/paper/2401.18059/) (2024)
- [Precise Zero-Shot Dense Retrieval without Relevance Labels](https://mlsanity.com/paper/2212.10496/) (2022)
- [A Dataset of Information-Seeking Questions and Answers Anchored in Research Papers](https://mlsanity.com/paper/2105.03011/) (2021)
- [HippoRAG: Neurobiologically Inspired Long-Term Memory for Large Language Models](https://mlsanity.com/paper/2405.14831/) (2024)
- [Nomic Embed: Training a Reproducible Long Context Text Embedder](https://mlsanity.com/paper/2402.01613/) (2024)
- [Document Expansion by Query Prediction](https://mlsanity.com/paper/1904.08375/) (2019)
- [Dense X Retrieval: What Retrieval Granularity Should We Use?](https://mlsanity.com/paper/2312.06648/) (2023)
- [Late Chunking: Contextual Chunk Embeddings Using Long-Context Embedding Models](https://mlsanity.com/paper/2409.04701/) (2024)
- [Financial Report Chunking for Effective Retrieval Augmented Generation](https://mlsanity.com/paper/2402.05131/) (2024)
- [Is Semantic Chunking Worth the Computational Cost?](https://mlsanity.com/paper/2410.13070/) (2024)
- [LumberChunker: Long-Form Narrative Document Segmentation](https://mlsanity.com/paper/2406.17526/) (2024)
- [Meta-Chunking: Learning Text Segmentation and Semantic Completion via Logical Perception](https://mlsanity.com/paper/2410.12788/) (2024)
- [FreshStack: Building Realistic Benchmarks for Evaluating Retrieval on Technical Documents](https://mlsanity.com/paper/2504.13128/) (2025)
- [A Systematic Framework for Enterprise Knowledge Retrieval: Leveraging LLM-Generated Metadata to Enhance RAG Systems](https://mlsanity.com/paper/2512.05411/) (2025)
- [Cross-Document Topic-Aligned Chunking for Retrieval-Augmented Generation](https://mlsanity.com/paper/2601.05265/) (2025)

## Abstract (arXiv)

Markdown carries structure a parser reads for free: headers, section paths, and block boundaries. Many RAG pipelines also spend LLM calls per chunk on generated metadata. We ask what one LLM call per chunk buys over that free structure. MDKeyChunker splits Markdown into header-led chunks without splitting any block; makes one LLM call per chunk for a title, summary, keywords, entities, questions, and a subtopic key, showing the model the keys already assigned in the document (a rolling key dictionary); and can merge same-key chunks. With qwen2.5:7b, we evaluate 79 Qasper questions over 30 papers and 73 FreshStack questions over 24 Laravel documentation files under BM25, two dense embedders, and hybrid fusion, following an analysis plan committed before results were computed. Evidence is matched only against source text, within a fixed token budget. Under hybrid retrieval, structural chunks beat 512-character windows on both datasets (Qasper +23.0 points, 95% CI [+12.8, +33.5]; FreshStack +5.1 [+1.4, +9.0]) and 256-token windows on Qasper (+12.7 [+5.3, +20.3]) but not on FreshStack (-2.6 [-6.4, +1.2]). Under the primary retrievers (hybrid, BM25), enrichment shows no planned-comparison difference from a free section-path prefix or from contextual retrieval; under hybrid retrieval the intervals exclude gains above 4.5 and 2.3 points on Qasper and 6.1 on FreshStack. Outside the planned comparisons, enrichment-style prefixes help BM25 on Qasper (exploratory) and mxbai on FreshStack (a secondary retriever). Rolling keys raise key reuse from 5.5% to 14.7%, but merging does not improve retrieval, and under BM25 on Qasper merging with rolling keys scores below merging without them (-6.0 [-13.1, -0.2]). Enrichment used about 1,000 input tokens per chunk; contextual retrieval 5,520 (Qasper) and 9,825 (FreshStack). The results of versions 1 and 2 are withdrawn.