TopoChunker: Topology-Aware Agentic Document Chunking Framework

TL;DR

TopoChunker is a novel framework that preserves document topology during chunking, improving retrieval quality and efficiency in RAG systems by explicitly modeling hierarchical dependencies.

Contribution

It introduces a dual-agent architecture for topology-aware document chunking, balancing structural fidelity with computational cost, and achieves state-of-the-art results on complex datasets.

Findings

Outperforms baseline by 8.0% in generation accuracy

Achieves 83.26% Recall@3

Reduces token overhead by 23.5%

Abstract

Current document chunking methods for Retrieval-Augmented Generation (RAG) typically linearize text. This forced linearization strips away intrinsic topological hierarchies, creating ``semantic fragmentation'' that degrades downstream retrieval quality. In this paper, we propose TopoChunker, an agentic framework that maps heterogeneous documents onto a Structured Intermediate Representation (SIR) to explicitly preserve cross-segment dependencies. To balance structural fidelity with computational cost, TopoChunker employs a dual-agent architecture. An Inspector Agent dynamically routes documents through cost-optimized extraction paths, while a Refiner Agent performs capacity auditing and topological context disambiguation to reconstruct hierarchical lineage. Evaluated on unstructured narratives (GutenQA) and complex reports (GovReport), TopoChunker demonstrates state-of-the-art performance. It outperforms the strongest LLM-based baseline by 8.0% in absolute generation accuracy and achieves an 83.26% Recall@3, while simultaneously reducing token overhead by 23.5%, offering a scalable approach for structure-aware RAG.