Dynamic Range-Filtering Approximate Nearest Neighbor Search
Summary: Define dynamic RFANNS: mixed stream of inserts and range-filtered ANN queries over attribute-constrained subsets; static RFANNS builds indexes by attribute order, so arbitrary arrival order breaks prior solutions. Propose the dynamic segment graph that compresses the O(|D|^2) HNSW graphs into a single index (lossless under conditions) with only linear-to-O(log |D|) expected edge additions per insert and practical heuristics, yielding much smaller indexes and superior empirical query performance. (summarized by gpt-5-mini on Feb 09 2026)
Incoming Non-self Citations Over Time
Authors
- 1. Zhencan Peng (Rutgers University)
- 2. Miao Qiao (University of Auckland)
- 3. Wenchao Zhou (Alibaba)
- 4. Feifei Li (Alibaba)
- 5. Dong Deng (Rutgers University)
BibTeX Citation
@article{peng_vldb25,
title = {{Dynamic Range-Filtering Approximate Nearest Neighbor Search}},
author = {Peng, Zhencan and Qiao, Miao and Zhou, Wenchao and Li, Feifei and Deng, Dong},
journal = {PVLDB},
series = {{VLDB} '25},
volume = {18},
number = {10},
pages = {3256--3268},
doi = {10.14778/3748191.3748193},
url = {https://doi.org/10.14778/3748191.3748193},
year = {2025}
}
Incoming Citations (Sorted by Pagerank)
Showing 8 of 8 citing papers.
Outgoing Citations (Sorted by Pagerank)
Showing 13 of 13 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.