OmniRetrieval: Unified Retrieval across Heterogeneous Knowledge Sources

Jinheon Baek, Soyeong Jeong, Sangwoo Park, Woongyeong Yeo, Minki Kang, Patara Trirat, Heejun Lee, Sung Ju Hwang

arXiv:2605.29250 · 2026-05-29 공개 · arXiv · PDF

benchmark-evaluation knowledge-graph graph-structured-data relational-databases omni-retrieval natural-language-query multi-source-retrieval heterogeneous-knowledge

Abstract

Real-world information needs require access to structurally diverse knowledge sources, from unstructured text and relational tables to knowledge graphs and property graphs. Existing retrievers, however, operate over one source at a time under a fixed query language, leaving the broader landscape of available knowledge fragmented behind incompatible interfaces. A natural attempt at unification would collapse these sources into a shared space, but this erases the structural affordances (such as schemas, ontologies, compositional operators) that give each source its expressive power. Effective retrieval over diverse knowledge, therefore, requires not homogenization but an overarching layer that meets each source on its own terms. To achieve this, we present OmniRetrieval, a framework that takes any natural-language query, identifies appropriate knowledge sources, and dispatches source-native queries to their native execution engines. Across an extensive benchmark spanning 13 datasets and 309 distinct knowledge bases over text, relational, and graph-structured sources, OmniRetrieval exceeds single-source baselines, demonstrating that it can serve as a general-purpose interface to the heterogeneous sources while preserving the structural distinctions that make each source valuable.

한국어 요약

📋 한 줄 요약

**[Unified Retrieval / 이질 지식 소스]** OmniRetrieval이 자연어 query를 textual·relational·KG·property graph 등 적합 source로 dispatch, 13 datasets·309 KB 벤치에서 source-native query 활용으로 single-source baseline 능가.

🎯 핵심 기여도

💡 핵심 아이디어

이질 지식 source 통합 retrieval은 source들을 공유 공간으로 homogenize하면 schema·ontology·operator의 structural 가치가 erase되므로, 각 source를 자체 terms로 맞이하는 overarching dispatching layer로 source-native query를 native engine에 보내야 한다.

🔬 기술적 접근법

📊 주요 결과

💭 의의 및 한계

**의의**: 이질 KB 통합 retrieval의 새 패러다임(homogenization 거부, native 보존), 실세계 정보 요구의 source 다양성 수용, 광범위 309 KB 벤치로 실증. **한계**: Source 식별·query 변환의 추가 복잡성·latency, source-native engine 의존(가용성·API 변화), 새 source 추가 시 dispatcher 학습 필요.

🚀 실용적 활용