Hybrid privacy-aware semantic search: SVD-truncated document geometry and CKKS-encrypted query reranking under a restricted threat model
Not on the map. Why?
Abstract (arXiv)
Semantic search creates an asymmetric disclosure problem: query embeddings may reveal user intent, while returning exact provider vectors distributes reusable representations. We evaluate a deliberately restricted hybrid design. A public corpus-fitted SVD basis and PQ index support client-side candidate selection; candidate IDs are disclosed, while the projected query is encrypted with CKKS. A block-SIMD kernel scores 100 plaintext provider vectors and returns one ciphertext. At 672 dimensions, median server time falls from 1689.1 to 224.5 ms and response size by a factor of 99.5; provider-only saturation reaches 25.99 requests/s with 16 workers. In a frozen post-exploratory revision-analysis subset, disjoint from validation and containing 3,235 canonical BEIR queries, five of six collections satisfy a +/-0.002 nDCG@10 equivalence rule between actual CKKS and plaintext reranking of the same shortlist, while ArguAna is inconclusive. Projection controls favor SVD over random and coordinate truncation. Leakage audits show that disclosed candidate sets are highly linkable and reproduce part of the exact neighbourhood, while public PQ reveals approximate corpus geometry. The design therefore conditionally hides numerical query slots, but does not provide semantic-query, document, unlinkability, circuit, or access-pattern privacy.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 34 of its 81 references that are in the map. First posted on arXiv 24 Jun 2026.
Builds on
34 of 81 references.
References · 34
- 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,297
- 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,170
- 2020Dense Passage Retrieval for Open-Domain Question Answeringcited by 3,824
- 2016Deep Learning with Differential Privacycited by 2,728
- 2017Billion-scale similarity search with GPUscited by 2,497
- 2016Membership Inference Attacks against Machine Learning Modelscited by 2,281
- 2023Retrieval-Augmented Generation for Large Language Models: A Surveycited by 2,212
- 2020Extracting Training Data from Large Language Modelscited by 2,189
- 2023Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflectioncited by 1,839
- 2022Text Embeddings by Weakly-Supervised Contrastive Pre-trainingcited by 1,253
- 2021BEIR: A Heterogenous Benchmark for Zero-shot Evaluation of Information Retrieval Modelscited by 1,252
- 2020ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERTcited by 1,178
22 more
- 2024M3-Embedding: Multi-Linguality, Multi-Functionality, Multi-Granularity Text Embeddings Through Self-Knowledge Distillationcited by 1,073
- 2016Efficient and robust approximate nearest neighbor search using Hierarchical Navigable Small World graphscited by 861
- 2021Large Dual Encoders Are Generalizable Retrieverscited by 456
- 2024Multilingual E5 Text Embeddings: A Technical Reportcited by 372
- 2020Information Leakage in Embedding Modelscited by 203
- 2024Corrective Retrieval Augmented Generationcited by 197
- 2018Gazelle: A Low Latency Framework for Secure Neural Network Inferencecited by 193
- 2023Text Embeddings Reveal (Almost) As Much As Textcited by 192
- 2024The Good and The Bad: Exploring Privacy Issues in Retrieval-Augmented Generation (RAG)cited by 159
- 2025LlamaFirewall: An open source guardrail system for building secure AI agentscited by 127
- 2023Sentence Embedding Leaks More Information than You Expect: Generative Embedding Inversion Attack to Recover the Whole Sentencecited by 81
- 2025Harnessing the Universal Geometry of Embeddingscited by 55
- 2024Transferable Embedding Inversion Attack: Uncovering Privacy Risks in Text Embeddings without Model Queriescited by 25
- 2019EVA: An Encrypted Vector Arithmetic Language and Compiler for Efficient Homomorphic Computationcited by 20
- 2019SANNS: Scaling Up Secure Approximate k-Nearest Neighbors Searchcited by 12
- 2021GALA: Greedy ComputAtion for Linear Algebra in Privacy-Preserved Neural Networkscited by 10
- 2021Intel HEXL: Accelerating Homomorphic Encryption with Intel AVX512-IFMA52cited by 8
- 2024Eguard: Defending LLM Embeddings Against Inversion Attacks via Text Mutual Information Optimizationcited by 8
- 2025Universal Zero-shot Embedding Inversioncited by 8
- 2025Privacy-Preserving LLM Embedding Transmission for End-Cloud Collaborationcited by 5
- 2026Zero2Text: Zero-Training Cross-Domain Inversion Attacks on Textual Embeddingscited by 4
- 2025Transform Before You Query: A Privacy-Preserving Approach for Vector Retrieval with Embedding Space Alignmentcited by 3
Built on by
Citing papers per year: 2026: 1.
1 paper.