在线社交网络的逻辑模型和并行查询

发布时间：2018-06-01 11:46

本文选题：复杂网络 + 平行算法　；参考：《复杂系统与复杂性科学》2013年02期

【摘要】：归纳出对在线社交网络研究具有挑战性的一些课题,介绍描述用户关系的逻辑模型(粉丝模型),提出逻辑关系寓意邻接矩阵(粉丝矩阵)。用此模型展示对微博平台Top-X信息查询的聚合-排序-删除算法。进一步应用映射和化简概念将上述Top-X信息查询算法扩展于并行计算环境,给出映射关注和化简粉丝在Ha-doop系统联机实现的算法。粉丝模型和相应的算法实现了对新浪微博74.7GB和Twitter的101GB实际数据的多种约束下信息查询和微博转发预测,特别是在Hadoop系统联机环境下,新方法的信息化简和计算性能明显提高。
[Abstract]:This paper summarizes some challenging topics in the research of online social networks, introduces the logical model (fan model) to describe the user relationship, and puts forward the logical relationship implication adjacency matrix (fan matrix). This model is used to demonstrate the aggregation-sort-delete algorithm for Top-X information query on Weibo platform. Furthermore, the concepts of mapping and simplification are applied to extend the above Top-X information query algorithm to the parallel computing environment, and the online implementation algorithm of mapping attention and simplification fans in Ha-doop system is given. The fan model and the corresponding algorithm have realized the information query and Weibo forwarding prediction for the actual 101GB data of Sina Weibo 74.7GB and Twitter under various constraints. Especially in the online Hadoop system, the information simplification and computational performance of the new method have been improved obviously.
【作者单位】：巴西利亚大学计算机系TransLab实验室;
【基金】：巴西科学技术发展委员会(CNPq,304058/2010-6,478039/2012-3)
【分类号】：TP393.0

【参考文献】