基于灰色关联分析的热门微博数据可视化研究

发布时间：2018-03-11 17:32

本文选题：关联规则　切入点：加权链接　出处：《郑州大学》2017年硕士论文　论文类型：学位论文

【摘要】：关联分析可视化是从数据中识别有效的、潜在的、有用的和可理解关系模式的一种非平凡过程。微博作为社交网络的一种传播介质,对社会舆论的传播有不可替代的作用。并且热门微博的出现使信息的传播更为迅速,是社会舆情的导向标,通过关联分析可视化技术,将微博之间以及微博与微博属性之间的相关联系进行直观的发掘,找出潜在的关联关系,对社会舆情的预防控制有重要的作用,因此对热门微博数据进行关联分析可视化的研究具有重要意义。关联分析挖掘算法是关联分析可视化过程中非常重要的部分,当前对关联分析可视化优化模型的研究处于初期阶段,现有的关联分析可视化方法也存在一些问题,将有效的数据挖掘模型与良好的可视化方法相结合,以便获得有价值的结果成为研究的重点。论文主要工作如下:1)为评价社交网络中微博博主影响力对微博传播的影响,结合微博在社交网络的传播特性以及信息传播的途径、方法,提出基于中心性加权链接强度的混合算法,该算法应用中心性加权链接强度,将启发式算法与贪心算法相结合,给出用户节点之间的影响力估计公式,从而计算每条微博传播者影响力的值,为评估热门微博属性的权重提供可靠依据。实验表明,相比较贪心算法、基于节点度的启发式算法,采用中心性加权链接强度算法在准确度上有明显的提升。2)针对微博数据经过无量纲化处理后,难以对其属性权重进行客观赋值的问题。提出一种基于灰度集合中多属性分析的方法。该方法结合微博博主的影响力与微博属性的关系,对多属性值进行定义和分类,建立挖掘过程参数调整策略,为选择关键属性值进行挖掘分析提供支持。3)针对现有关联分析可视化算法存在一些不足,将社交网络数据作为应用背景,提出基于灰色关联分析的可视化算法。该算法利用灰色关联分析理论,将热门微博中不同属性的数据区分,解决关联规则数量较多时传统关联分析方法面临的可视化界面紊乱、产生歧义等问题。实验结果表明,改进后的挖掘算法具有较好的性能,所提出的可视化形式和已有成果相比具有良好的展现效果。
[Abstract]:Association analysis visualization is a nontrivial process of identifying valid, potential, useful and understandable relational patterns from data. It has an irreplaceable effect on the dissemination of public opinion. And the emergence of hot Weibo makes the dissemination of information more rapid. It is the guide of social public opinion, and through the visual technology of association analysis, It is important for the prevention and control of social public opinion to explore directly the correlation between Weibo and the attributes of Weibo and to find out the potential relationship. Therefore, it is of great significance to study the association analysis visualization of hot Weibo data. The association analysis mining algorithm is a very important part in the process of association analysis visualization. At present, the research on the visualization optimization model of association analysis is in the initial stage, and the existing visualization methods of association analysis also have some problems, which combine the effective data mining model with the good visualization method. In order to obtain valuable results, the main work of this paper is as follows: 1) in order to evaluate the influence of Weibo's influence on the transmission of Weibo in social networks, combining the communication characteristics and the ways of information dissemination in social networks, Methods A hybrid algorithm based on the central weighted link strength is proposed. The heuristic algorithm is combined with the greedy algorithm and the influence estimation formula between the user nodes is given by applying the central weighted link strength. Thus the value of each Weibo communicator's influence is calculated, which provides a reliable basis for evaluating the weight of the popular Weibo attribute. The experiment shows that compared with greedy algorithm, the heuristic algorithm based on node degree, The centrality weighted link strength algorithm is used to improve the accuracy obviously. 2) after the data of Weibo is dimensionless, It is difficult to assign the attribute weight objectively. This paper presents a method based on multi-attribute analysis in gray level set. This method defines and classifies the multi-attribute value based on the relationship between Weibo's influence and Weibo attribute. A mining process parameter adjustment strategy is established to support mining analysis by selecting key attribute values. 3) aiming at the shortcomings of the existing visualization algorithms for association analysis, the social network data is used as the application background. This paper presents a visualization algorithm based on grey association analysis, which uses the theory of grey correlation analysis to distinguish the data of different attributes in the popular Weibo, and solves the visual interface disorder faced by the traditional association analysis method when the number of association rules is large. The experimental results show that the improved mining algorithm has better performance and the proposed visualization has a good performance compared with the existing results.
【学位授予单位】：郑州大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13

【相似文献】