图数据流的模型、算法和系统  

Graph stream: model, algorithm and system

作　　者：李友焕 邹磊[1] LI Youhuan;ZOU Lei(Institute of Computer Science ＆ Technology of Peking University,Beijing 100080,China)

机构地区：[1]北京大学计算机科学技术研究所,北京100080

出　　处：《大数据》2018年第4期44-55,共12页Big Data Research

摘　　要：在应用数据高速增长的场景下,已有的静态图计算的模型和方法难以应对数据高速更新的挑战,图数据流模型应运而生。首先讨论当前大规模复杂数据流的产生及其管理需求,分析静态图模型以及已有数据流算法、系统在应对这一数据流场景的固有缺陷,阐述图数据流模型产生的重要背景。然后通过总结分析早期图的流式计算以及已有的少量图数据流的研究工作,给出图数据流模型的一般定义。最后,从方法和问题两个角度探讨图数据流的研究前景,并简要介绍图数据流管理系统相关技术架构。In the scenario where data of real-world applications is in high-speed growth, existing methods for static graph computation are hard to approach the challenges from the rapidly updated data, and graph stream model arise at the historic moment. The inherent defects of static graph model and exiting data stream algorithms/systems over high-speed graph-structured data were discussed, and then the formal definition of graph stream combining with previous similar models was presented. Some promising research problems and applications over graph stream were probed, and then the possible requirements and technique issues on graph stream management system（GSMS） were looked into the future.

关 键 词：图模型 数据流系统 图数据流 数据管理系统 

分 类 号：TP315[自动化与计算机技术—计算机软件与理论]