Trans-RAG: Query-Centric Vector Transformation for Secure Cross-Organizational Retrieval
2026-04-10 • Cryptography and Security
Cryptography and SecurityInformation Retrieval
AI summaryⓘ
The authors address challenges in sharing knowledge securely and efficiently between organizations using Retrieval Augmented Generation (RAG) systems. They propose Trans-RAG, which keeps each organization’s information in separate, mathematically isolated spaces to avoid exposing data during retrieval. Their method uses a transformation called vector2Trans that allows queries to adapt dynamically to each organization's data space without costly decryption steps. Tests show that this approach maintains strong security while only slightly reducing accuracy and improving efficiency compared to traditional encryption methods.
Retrieval Augmented Generationvector spacesemantic spaceencryptionhomomorphic encryptionfederated learningquery transformationlarge language modelsinformation retrievalsecurity
Authors
Yu Liu, Kun Peng, Wenxiao Zhang, Fangfang Yuan, Cong Cao, Wenxuan Lu, Yanbing Liu
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.