ContextRAG: Extraction-Free Hierarchical Graph Construction for Retrieval-Augmented Generation

TL;DR

ContextRAG introduces a novel graph construction method for retrieval-augmented generation that eliminates reliance on large language models for entity and relation extraction, reducing costs and maintaining competitive performance.

Contribution

It presents a new extraction-free hierarchical graph construction approach using fuzzy concept graphs, significantly lowering indexing costs compared to traditional LLM-dependent methods.

Findings

ContextRAG reduces indexing calls from 870 to 30 on a 130-task dataset.

It achieves 33.6% F1 score on a multi-task benchmark.

Fuzzy concept graph nodes improve retrieval effectiveness.

Abstract

Graph-structured retrieval-augmented generation (RAG) systems can improve answer quality on multi-hop questions, but many current systems rely on large language models (LLMs) to extract entities, relations, and summaries during indexing. These calls add token and wall-clock costs that grow with corpus size. We present ContextRAG, a graph RAG system whose graph topology is constructed without LLM-based entity or relation extraction. ContextRAG derives a fuzzy concept graph over chunk embeddings using residual-quantization k-means and Formal Concept Analysis with Lukasiewicz residuated logic. Bridge-like and meet-derived context nodes are induced by soft fuzzy join and meet operations, rather than by LLM-written graph edges. On a 130-task UltraDomain subset, ContextRAG builds its index with 30 LLM calls and 22,073 tokens. In contrast, a local HiRAG reproduction stress test required 870 indexing calls and 3.54M tokens on a 20-task subset before failing during graph construction; linear extrapolation to 130 tasks implies over 23M indexing tokens. ContextRAG obtains 33.6% F1 overall and 36.8% F1 on multi-hop tasks. An activation analysis shows that queries retrieving at least one lattice-derived node in the top five achieve +3.9 percentage points F1 over queries that do not; this association is diagnostic rather than causal.