Geometric graphs improve multi-step document retrieval and generation
Beyond Similarity through Zero-Token Geometric Graphs for Multi-Hop RAG
Information Retrieval
Summary
Finding information from multiple connected documents is tricky because similar documents tend to repeat the same facts, while others that add helpful new details can be missed. The authors developed a new way to build a network of documents without relying on expensive language model analysis. Their method uses geometry to measure how documents connect in a way that balances similarity and novelty. This approach helps find more useful supporting documents for complex questions, improving accuracy while saving computing costs.
What this means in practice
- •For search engine engineers: Enhance retrieval systems for complex queries by discovering novel relevant documents efficiently without costly language model-based processing.
- •For customer support platform developers: Build better multi-document answer generation tools that find supporting evidence faster and more accurately from large knowledge bases.
Authors
Zeliang Li, Xiaofen Xing, Kailing Guo, Xiangmin Xu
Abstract
Multi-hop retrieval-augmented generation (RAG) requires evidence that remains relevant to a query while introducing enough novelty to bridge semantic gaps. Dense retrieval tends to concentrate on semantically similar documents, whereas graph-based alternatives often depend on costly Large Language Model (LLM) entity extraction and may propagate through noisy connections. We introduce Geometric Gain Graph RAG (G$^3$RAG), a document-only framework whose offline graph construction uses no LLM calls or generated tokens. G$^3$RAG assigns each edge a geometric gain score, $\cosθ\cdot \sinθ$, that jointly captures directional consistency and orthogonality between document representations. A density-aware topological penalty suppresses highly connected hubs, while single-step controlled diffusion expands from filtered query seeds toward complementary evidence. We evaluate G$^3$RAG on MusiQue, 2WikiMultiHopQA, and HotpotQA using Nv-embed-v2 and Qwen3-8B-embed. G$^3$RAG obtains the best average F1 and answer-document hit rate among the evaluated graph-based baselines in both embedding settings, with gains of up to 4.26 F1 points in average performance and 5.76 points on MusiQue. It also removes the graph-construction token cost incurred by entity-based graph methods. These results show that geometric structure can support efficient multi-hop evidence discovery without LLM-based graph construction. Code is available at https://anonymous.4open.science/r/G3RAG-99D9/