SAGE: Structure Aware Graph Expansion for Retrieval of Heterogeneous Data

TL;DR

SAGE introduces a structure-aware graph expansion method for retrieval in heterogeneous data, enhancing multi-modal evidence retrieval by combining offline graph construction with online neighbor expansion and filtering.

Contribution

The paper presents a novel framework that constructs a chunk-level graph offline and performs online multi-hop retrieval, improving retrieval recall over existing methods.

Findings

SAGE improves retrieval recall by 5.7 points on OTT-QA.

SAGE improves retrieval recall by 8.5 points on STaRK.

The method effectively combines offline graph construction with online neighbor expansion.

Abstract

Retrieval-augmented question answering over heterogeneous corpora requires connected evidence across text, tables, and graph nodes. While entity-level knowledge graphs support structured access, they are costly to construct and maintain, and inefficient to traverse at query time. In contrast, standard retriever-reader pipelines use flat similarity search over independently chunked text, missing multi-hop evidence chains across modalities. We propose SAGE (Structure Aware Graph Expansion) framework that (i) constructs a chunk-level graph offline using metadata-driven similarities with percentile-based pruning, and (ii) performs online retrieval by running an initial baseline retriever to obtain k seed chunks, expanding first-hop neighbors, and then filtering the neighbors using dense+sparse retrieval, selecting k' additional chunks. We instantiate the initial retriever using hybrid dense+sparse retrieval for implicit cross-modal corpora and SPARK (Structure Aware Planning Agent for Retrieval over Knowledge Graphs) an agentic retriever for explicit schema graphs. On OTT-QA and STaRK, SAGE improves retrieval recall by 5.7 and 8.5 points over baselines.