Trans-RAG: Query-Centric Vector Transformation for Secure Cross-Organizational Retrieval
Yu Liu, Kun Peng, Wenxiao Zhang, Fangfang Yuan, Cong Cao + 2 more
TLDR
Trans-RAG introduces query-centric vector transformation for secure, efficient cross-organizational RAG, eliminating decryption overhead.
Key contributions
- Introduces Trans-RAG, a novel vector space language paradigm for secure, isolated cross-organizational knowledge.
- Employs `vector2Trans`, a multi-stage technique for query-centric vector space transformation.
- Achieves high security with near-orthogonal vector spaces (89.90° separation, 99.81% isolation).
- Demonstrates minimal accuracy loss (3.5% nDCG@10) and superior efficiency over homomorphic encryption.
Why it matters
This paper addresses the critical challenge of deploying RAG systems securely and efficiently across organizational boundaries. By introducing a novel vector transformation approach, Trans-RAG overcomes limitations of existing encryption and federated methods. It offers a practical solution for secure knowledge sharing without compromising performance.
Original Abstract
Retrieval Augmented Generation (RAG) systems deployed across organizational boundaries face fundamental tensions between security, accuracy, and efficiency. Current encryption methods expose plaintext during decryption, while federated architectures prevent resource integration and incur substantial overhead. We introduce Trans-RAG, implementing a novel vector space language paradigm where each organization's knowledge exists in a mathematically isolated semantic space. At the core lies vector2Trans, a multi-stage transformation technique that enables queries to dynamically "speak" each organization's vector space "language" through query-centric transformations, eliminating decryption overhead while maintaining native retrieval efficiency. Security evaluations demonstrate near-orthogonal vector spaces with 89.90° angular separation and 99.81% isolation rates. Experiments across 8 retrievers, 3 datasets, and 3 LLMs show minimal accuracy degradation (3.5% decrease in nDCG@10) and significant efficiency improvements over homomorphic encryption.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.