HHP: A Hybrid Partitioner for Large-Scale Hypergraph

Junlin Shang, Zhenyu Zhang, Wenwen Qu, Xiaoling Wang*

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Hypergraphs have diverse applications in building n-ary relationships, and their partitioning is crucial for distributed systems that utilize hypergraph-structured data. Hypergraph partitioners can be classified into offline and online strategies. Offline strategies deliver high-quality results but demand significant time and memory resources. Conversely, online strategies require fewer resources but may produce lower-quality partitions. In this paper, we introduce a novel hypergraph partitioner, the Hybrid Hypergraph Partitioner (HHP), designed to achieve high-quality partitioning with limited resources. HHP adapts resource consumption by splitting the hypergraph into two sub-hypergraphs and completing the partitioning in two steps. First, HHP applies offline partitioning to one sub-hypergraph, followed by an online strategy that uses the stateful information from the first step to partition the remaining sub-hypergraph. Our evaluation of large-scale hypergraphs shows that HHP achieves superior partitioning quality while reducing both time and memory consumption.

Original languageEnglish
Title of host publicationAdvanced Data Mining and Applications - 20th International Conference, ADMA 2024, Proceedings
EditorsQuan Z. Sheng, Xuyun Zhang, Jia Wu, Congbo Ma, Gill Dobbie, Jing Jiang, Wei Emma Zhang, Yannis Manolopoulos, Wathiq Mansoor
PublisherSpringer Science and Business Media Deutschland GmbH
Pages114-129
Number of pages16
ISBN (Print)9789819608201
DOIs
StatePublished - 2025
Event20th International Conference on Advanced Data Mining Applications, ADMA 2024 - Sydney, Australia
Duration: 3 Dec 20245 Dec 2024

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume15389 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference20th International Conference on Advanced Data Mining Applications, ADMA 2024
Country/TerritoryAustralia
CitySydney
Period3/12/245/12/24

Keywords

  • Hybrid partitioning
  • Hypergraph partitioning
  • Neighborhood expansion
  • Streaming

Fingerprint

Dive into the research topics of 'HHP: A Hybrid Partitioner for Large-Scale Hypergraph'. Together they form a unique fingerprint.

Cite this