LGP-SA:Graph partition algorithm based on simulated annealing in large-scale graph processing

Distributed computing for large-scale graph data need to partition the graph firstly. The current methods of large-scale graph partitioning is to reduce the edge cut in order to lessen communication overhead by using vertex transfer strategies,but easily to fall into local optimum. Simulated anneali...

Full description

Saved in:
Bibliographic Details
Main Authors: Jinfeng XU, Yihong DONG, Shiyi WANG, Xianmang HE, Huahui CHEN
Format: Article
Language:zho
Published: Beijing Xintong Media Co., Ltd 2016-02-01
Series:Dianxin kexue
Subjects:
Online Access:http://www.telecomsci.com/zh/article/doi/10.3969/j.issn.1000-0801.2016.02.012/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1841529765799919616
author Jinfeng XU
Yihong DONG
Shiyi WANG
Xianmang HE
Huahui CHEN
author_facet Jinfeng XU
Yihong DONG
Shiyi WANG
Xianmang HE
Huahui CHEN
author_sort Jinfeng XU
collection DOAJ
description Distributed computing for large-scale graph data need to partition the graph firstly. The current methods of large-scale graph partitioning is to reduce the edge cut in order to lessen communication overhead by using vertex transfer strategies,but easily to fall into local optimum. Simulated annealing has a great probability to avoid the trap of local optimum and prevent vertices from invalid transfer which was introduced to transfer vertices. This method decreased communication overhead greatly. Comparative experiments show that the proposed algorithm has made a great improvement in reducing edge cut rates in large scale graph partition field. PageRank algorithm was also used to verify the effectiveness and feasibility of this method.
format Article
id doaj-art-1084239c9c38457fbbd7247a5c217288
institution Kabale University
issn 1000-0801
language zho
publishDate 2016-02-01
publisher Beijing Xintong Media Co., Ltd
record_format Article
series Dianxin kexue
spelling doaj-art-1084239c9c38457fbbd7247a5c2172882025-01-15T03:15:20ZzhoBeijing Xintong Media Co., LtdDianxin kexue1000-08012016-02-0132839159610220LGP-SA:Graph partition algorithm based on simulated annealing in large-scale graph processingJinfeng XUYihong DONGShiyi WANGXianmang HEHuahui CHENDistributed computing for large-scale graph data need to partition the graph firstly. The current methods of large-scale graph partitioning is to reduce the edge cut in order to lessen communication overhead by using vertex transfer strategies,but easily to fall into local optimum. Simulated annealing has a great probability to avoid the trap of local optimum and prevent vertices from invalid transfer which was introduced to transfer vertices. This method decreased communication overhead greatly. Comparative experiments show that the proposed algorithm has made a great improvement in reducing edge cut rates in large scale graph partition field. PageRank algorithm was also used to verify the effectiveness and feasibility of this method.http://www.telecomsci.com/zh/article/doi/10.3969/j.issn.1000-0801.2016.02.012/graph partitionGiraphsimulated annealinglarge-scale graphBSP
spellingShingle Jinfeng XU
Yihong DONG
Shiyi WANG
Xianmang HE
Huahui CHEN
LGP-SA:Graph partition algorithm based on simulated annealing in large-scale graph processing
Dianxin kexue
graph partition
Giraph
simulated annealing
large-scale graph
BSP
title LGP-SA:Graph partition algorithm based on simulated annealing in large-scale graph processing
title_full LGP-SA:Graph partition algorithm based on simulated annealing in large-scale graph processing
title_fullStr LGP-SA:Graph partition algorithm based on simulated annealing in large-scale graph processing
title_full_unstemmed LGP-SA:Graph partition algorithm based on simulated annealing in large-scale graph processing
title_short LGP-SA:Graph partition algorithm based on simulated annealing in large-scale graph processing
title_sort lgp sa graph partition algorithm based on simulated annealing in large scale graph processing
topic graph partition
Giraph
simulated annealing
large-scale graph
BSP
url http://www.telecomsci.com/zh/article/doi/10.3969/j.issn.1000-0801.2016.02.012/
work_keys_str_mv AT jinfengxu lgpsagraphpartitionalgorithmbasedonsimulatedannealinginlargescalegraphprocessing
AT yihongdong lgpsagraphpartitionalgorithmbasedonsimulatedannealinginlargescalegraphprocessing
AT shiyiwang lgpsagraphpartitionalgorithmbasedonsimulatedannealinginlargescalegraphprocessing
AT xianmanghe lgpsagraphpartitionalgorithmbasedonsimulatedannealinginlargescalegraphprocessing
AT huahuichen lgpsagraphpartitionalgorithmbasedonsimulatedannealinginlargescalegraphprocessing