Don't Be a Pot Stirrer! Authorized Vector Data Retrieval via Access-Aware Indexing

2026-05-02

AI Summary

This work addresses the challenge of efficiently executing top-k approximate nearest neighbor (ANN) queries that return only authorized vectors in role-based access-controlled vector databases, where storage overhead and query efficiency present a fundamental trade-off. The authors propose Veda and EffVeda, two novel indexing strategies that introduce, for the first time, an access-aware lattice structure to partition data by role combinations. Under a user-specified storage budget, these methods aggregate co-accessed blocks through replication and merging, indexing large blocks with HNSW while retaining linear scan for small ones. For each role, a minimal covering node query plan is constructed, integrating pure-authorized-node prioritized search with non-pure node pruning to drastically reduce irrelevant retrievals. Experiments demonstrate that the proposed approach significantly outperforms both global indexing and an idealized full-authorization index, achieving superior recall and query efficiency while maintaining controlled storage overhead.

Abstract

Vector databases increasingly enforce role-based access control: each top-k approximate nearest neighbor query must return only vectors the querying role is authorized to access. Two extremes bracket the design space. A single global index avoids data duplication but wastes search effort on unauthorized vectors and degrades recall, while an oracle index, built with all authorized vectors of the query roles, searches only authorized vectors but duplicates every shared vector between roles or queries. We present Veda and its efficient variant EffVeda, two indexing strategies built on an access-aware lattice to address access control in vector databases. The methods first partitions the dataset into disjoint data blocks by role combination, then leverage the structure of the access-aware lattice to apply copy and merge operations to group co-accessed blocks under a user-specified storage budget. Large nodes in the lattice are then indexed with HNSW, while small nodes are retained for linear scan. For each role, our methods construct a query plan that selects the minimal set of nodes that covers the role's authorized data. At query time, coordinated search first queries pure (authorized-only) nodes to populate a global top-k heap. The resulting distance bound then prunes exploration on impure nodes, avoiding the inflated search that independent per-index execution would require.

Problem

• vector databases
• role-based access control
• approximate nearest neighbor
• authorized data retrieval
• access-aware indexing

Innovation

• access-aware indexing
• vector database
• role-based access control
• approximate nearest neighbor search
• lattice-based data partitioning