Federated Vector Retrieval across Heterogeneous VectorDBs
cs.DB, cs.IR
Submitted: 2026-11-01
Updated: 2026-11-13
Comments: Accepted for publication at VLDB 2027. The manuscript contains 13 figures
Code: https://github.com/yoon123seul/HVMap
License: http://creativecommons.org/licenses/by/4.0/
The gist: Vector databases are now core infrastructure for AI retrieval systems, yet many real deployments must serve queries over multiple independently constructed vectorDBs built with different embedding…
Abstract
Vector databases are now core infrastructure for AI retrieval systems, yet many real deployments must serve queries over multiple independently constructed vectorDBs built with different embedding models. This creates a systems problem of federated retrieval over heterogeneous vectorDBs: vectors and similarity scores are not directly comparable across embedding spaces, while fully resolving this incompatibility through corpus-wide re-embedding and re-indexing is often operationally prohibitive. Existing methods therefore face a difficult trade-off between retrieval quality, construction cost, and query throughput. We present HVMap, a lightweight compatibility layer for federated retrieval across heterogeneous vectorDBs. Instead of rebuilding databases into a unified space, HVMap learns bidirectional mappings between embedding spaces using a fuzzy contrastive loss, enabling calibrated cross-database scoring without corpus migration or index reconstruction. HVMap is system-agnostic and further scales to multi-vectorDB settings while preserving ranking fidelity through a hub-based architecture. We evaluate HVMap in open-domain, cross-domain, and scale-out settings, and also construct a new benchmark for heterogeneous cross-domain federation. Across these settings, HVMap consistently achieves a better quality--cost--throughput trade-off than baselines, achieving up to 3.59X higher nDCG@10, up to 2.01X higher query throughput, and construction time as low as 0.3% of baseline methods. These results show that HVMap makes heterogeneous vectorDB federation practical at scale.
Sources
- RAG vs Fine-tuning: Pipelines, Tradeoffs, and a Case Study on Agriculture
- Improving language models by retrieving from trillions of tokens
- Language-agnostic BERT Sentence Embedding
- Characterizing Diseases from Unstructured Text: A Vocabulary Driven Word2vec Approach
- Efficient Federated Search for Retrieval-Augmented Generation using Lightweight Routing
- Distilling Knowledge from Reader to Retriever for Question Answering
- Generalization through Memorization: Nearest Neighbor Language Models
- NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models
- Gecko: Versatile Text Embeddings Distilled from Large Language Models
- Situating Sentence Embedders with Nearest Neighbor Overlap
- RoBERTa: A Robustly Optimized BERT Pretraining Approach
- PRIVEE: Privacy-Preserving Vertical Federated Learning Against Feature Inference Attacks
- Sentence-T5: Scalable Sentence Encoders from Pre-trained Text-to-Text Models
- GPT-4o System Card
- DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
- SsciBERT: A Pre-trained Language Model for Social Science Texts
- REPLUG: Retrieval-Augmented Black-Box Language Models
- Do We Need Domain-Specific Embedding Models? An Empirical Investigation
- SciFact-Open: Towards open-domain scientific claim verification
- From Word Vectors to Multimodal Embeddings: Techniques, Applications, and Future Directions For Large Language Models
Related papers
- Beyond Overlap: Hierarchical Graph Abstraction for Out-of-Core Vector Search
- EMA: Approximate Nearest Neighbor Search with General Attribute Filtering and Dynamic Updates
- Open Biomedical Knowledge Graphs at Scale: Construction, Federation, and AI Agent Access with Samyama Graph Database
- Samyama: A Unified Graph-Vector Database with In-Database Optimization, Agentic Enrichment, and Hardware Acceleration
- GenTUS: Table Union Search via Constrained Generation of Semantic Table Identifiers
- Common-Neighbor-Count-Based Representative Possible World Finding on Uncertain Graphs