首页 | 本学科首页   官方微博 | 高级检索  
     

路径-维度GraphOLAP大规模多维网络并行分析框架
引用本文:张子兴,吴斌,吴心宇,张有杰,孙思瑞,彭程程,刘昱彤.路径-维度GraphOLAP大规模多维网络并行分析框架[J].软件学报,2018,29(3):545-568.
作者姓名:张子兴  吴斌  吴心宇  张有杰  孙思瑞  彭程程  刘昱彤
作者单位:北京邮电大学 智能通信软件与多媒体北京市重点实验室, 北京 100876,北京邮电大学 智能通信软件与多媒体北京市重点实验室, 北京 100876,北京邮电大学 智能通信软件与多媒体北京市重点实验室, 北京 100876,北京邮电大学 智能通信软件与多媒体北京市重点实验室, 北京 100876,北京邮电大学 智能通信软件与多媒体北京市重点实验室, 北京 100876,北京邮电大学 智能通信软件与多媒体北京市重点实验室, 北京 100876,北京邮电大学 智能通信软件与多媒体北京市重点实验室, 北京 100876
基金项目:国家重点基础研究发展规划(973计划)项目(2013CB329606);国家自然科学基金(61772082)
摘    要:现实生活中大量数据都可以使用多维网络进行建模,如何更好地对多维网络进行分析至今仍是研究人员关注的重点.OLAP(联机分析处理)技术已被证实是对多维关系数据进行分析的有效工具,但应用OLAP技术管理和分析多维网络数据以支持有效决策仍旧是一项巨大的挑战.本文设计并提出了一种新的图立方体模型:路径-维度立方体,并针对提出的立方体模型将物化过程划分为关系路径物化与关联维度物化两部分,分别提出了物化策略并基于Spark框架设计了相关算法;在此基础上,我们针对网络数据设计并细化了相关的GraphOLAP(图联机分析处理)操作,丰富了框架的分析角度,提高了对多维网络的分析能力;最后,在Spark上实现了相关算法,通过对多个真实应用场景中的数据构建多维网络,在分析框架上进行了分析,实验表明我们提出的图立方体模型和物化算法具有一定有效性和可扩展性.

关 键 词:图立方体  立方体物化  关系路径  图联机分析处理
收稿时间:2017/7/31 0:00:00
修稿时间:2017/9/5 0:00:00

P&D GraphOLAP: Parallel Framework for Large-Scale Multidimensional Network Analysis
ZHANG Zi-Xing,WU Bin,WU Xin-Yu,ZHANG You-Jie,SUN Si-Rui,PENG Cheng-Cheng and LIU Yu-Tong.P&D GraphOLAP: Parallel Framework for Large-Scale Multidimensional Network Analysis[J].Journal of Software,2018,29(3):545-568.
Authors:ZHANG Zi-Xing  WU Bin  WU Xin-Yu  ZHANG You-Jie  SUN Si-Rui  PENG Cheng-Cheng and LIU Yu-Tong
Affiliation:Beijing Key Laboratory of Intelligent Telecommunications Software and Multimedia, Beijing University of Posts and Telecommunications, Beijing 100876, China,Beijing Key Laboratory of Intelligent Telecommunications Software and Multimedia, Beijing University of Posts and Telecommunications, Beijing 100876, China,Beijing Key Laboratory of Intelligent Telecommunications Software and Multimedia, Beijing University of Posts and Telecommunications, Beijing 100876, China,Beijing Key Laboratory of Intelligent Telecommunications Software and Multimedia, Beijing University of Posts and Telecommunications, Beijing 100876, China,Beijing Key Laboratory of Intelligent Telecommunications Software and Multimedia, Beijing University of Posts and Telecommunications, Beijing 100876, China,Beijing Key Laboratory of Intelligent Telecommunications Software and Multimedia, Beijing University of Posts and Telecommunications, Beijing 100876, China and Beijing Key Laboratory of Intelligent Telecommunications Software and Multimedia, Beijing University of Posts and Telecommunications, Beijing 100876, China
Abstract:Most data in real life can be described as multidimensional networks. How to process the analysis on multidimensional networks from multiple views and multiple granularities is still the focus of current research. Meanwhile, OLAP (Online Analytical Processing) technology has been proven to be an effective tool on relational data. However, it is an enormous challenge to manage and analyse multidimensional heterogeneous networks via OLAP technology to support effective decision making. In this paper, we propose a P&D (Path and Dimension) Graph Cube model. Based on this model, we divide the graph cube materialization into two parts described as Path related Materialization and Dimension related Materialization, and design the materialization algorithm implemented on Spark. And we also design and refine some GraphOLAP operations to improve the ability of analyzing multidimensional networks. Finally, we implement the algorithms on Spark and construct the multidimensional networks through real datasets. Then we analyse these networks using the framework. The results of experiments validate the effectiveness and scalability of P&D Graph Cube Model and the materialization algorithms.
Keywords:graph data cube  materialization  relation path  GraphOLAP
点击此处可从《软件学报》浏览原始摘要信息
点击此处可从《软件学报》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号