大规模交往数据集的凝聚子群分析研究

【作者】 万怀宇；

【导师】 黄厚宽；

【作者基本信息】 北京交通大学 ， 计算机应用技术， 2007， 硕士

【摘要】 随着近年来通信技术的蓬勃发展,通信领域内的各个企业都积累了大量的客户交往数据集。客户交往数据集是一个信息量非常庞大的数据资源,如果能够采用有效的方法和技术对其进行凝聚子群分析,我们将会从中获得许多有价值的规律和知识。本文提出了一种全新的进行交往数据集凝聚子群分析的方法——社会网络分析方法,它从社会学的角度,对由用户和用户间的通信关系所构成的交往网络进行了分析,从而在企业决策支持和犯罪侦查等多种领域提供服务。社会网络分析方法是一种量化的社会学分析方法。它将社会行动者映射为图的节点,社会行动者之间的关系映射为图中的边,然后利用图论的相关知识来解决社会网络的问题。本文设计了一个基于社会网络分析方法的大规模交往数据集凝聚子群分析模型,并对模型中的数据预处理、基本数据结构的定义、数据接口定义、关键图算法的实现、社会网络分析中心性指标集的计算、凝聚子群分析以及数据可视化等各个功能模块提出了具体的解决方案,并完成了其中部分模块的编码实现以及实验验证。更多还原

【Abstract】 Recent years, every corporation in the communication realm has collected a large number of communication dataset of consumers along with the development of communication technology. The communication dataset is a very valuable resource which contains much useful information. If we can use effective techniques and methods to analyze the dataset, we will get much valuable knowledge.The thesis adopts an absolutely new method– social network analysis method– to analyze the communication dataset. We use the new method to analyze the communication network composed of consumers and communication relationships between them from the point of view of sociology. And this will provide much assistance in many fields such as decision support of enterprise and crime investigation.Social network analysis method is a quantitative sociology method. It takes a social actor as a vertex of a graph, and takes the relationship between two actors as an edge of a graph, and then use the algorithm of graph theory to resolve the problems of social network.The thesis designs a cohesive subgroup analysis model of large-scale communication dataset based on social network analysis method. And it finishes many substantial resolutions to the modules of data pretreatment, define of basic data structures, define of data interface, implement of pivotal graph algorithm, calculation of social network analysis centrality measures, cohesive subgroup analysis and data visualization. Many of the modules have been implemented by coding and validated by experimenting.更多还原