社交网络中节点影响力评估算法研究
发布时间：2019-04-16 19:41
【摘要】：随着Web2.0的发展,在线社交网络已逐渐成为互联网中最流行的交流平台。社交网络中的用户可以相互结识并分享传播各种信息。社交网络中的海量用户及用户发布的信息具有巨大的商业价值和研究价值,这吸引了越来越多的研究者从不同的角度研究社交网络。评估社交网络节点影响力是社交网络研究的热点方向之一。通过评估社交网络节点影响力可以发现社交网络中的高影响力人物,对广告投放、信息管控和用户行为分析具有重要意义。 本文主要研究社交网络节点影响力评估方法,以更有效地发现社交网络中的高影响力人物。本文在对当前评估节点影响力相关方法进行分析研究的基础上,提出了一种新的社交网络节点影响力评估方法,并通过实验对该方法评估节点影响力的效果进行了验证和分析。 针对当前有关社交网络中节点影响力的研究很少关注影响力在不同主题下的差异以及用户在不同社区中扩散信息的能力的问题,本文提出了一种新的评估基于主题的节点影响力的方法。该方法综合考虑了用户间传播信息的可能性和信息被传播后可能继续传播的范围,结合了社交网络中用户间的关系、用户的发布转发行为和用户发布的信息内容等因素分析用户影响力,通过用户间亲密度和社交圈差异度来评估用户在不同主题下的影响力排名。 本文使用新浪微博中的用户和微博信息作为实验数据集,对根据本文方法得到的用户影响力评估结果进行了分析。实验表明,本文提出的方法能够得到不同主题下的用户影响力排名,不同主题的用户影响力排名间的相关度较低,高影响力用户在多个社区中具有更高的信息传播能力,排名越高的用户的信息在不同社区间的传播越均匀。
[Abstract]:With the development of Web2.0, online social network has gradually become the most popular communication platform in the Internet. Users in social networks can get to know each other and share and spread information. The mass of users in social networks and the information published by users have great business value and research value, which attracts more and more researchers to study social networks from different perspectives. Evaluating the influence of social network nodes is one of the hot topics in the research of social networks. By evaluating the influence of social network nodes, we can find high-impact people in social networks, which is of great significance to advertisement delivery, information control and user behavior analysis. In this paper, we mainly study the evaluation method of social network node influence, in order to find the high-impact person in the social network more effectively. On the basis of the analysis and research of the current methods of evaluating node influence, this paper proposes a new method of evaluating node influence of social network, and validates and analyzes the effect of evaluating node influence by experiments. Current research on node influence in social networks has paid little attention to the differences in influence under different topics and the ability of users to spread information across different communities. In this paper, a new method for evaluating the influence of topic-based nodes is proposed. The method takes into account the possibility of spreading information among users and the extent to which information may continue to spread after it is transmitted, and combines the relationships between users in social networks. The factors such as user's posting and forwarding behavior and the information content published by users are analyzed. The influence ranking of users under different topics is evaluated by the degree of affinity density and social circle difference among users. In this paper, the users and Weibo information in Sina Weibo are used as experimental data sets, and the results of user impact evaluation based on this method are analyzed. Experiments show that the method proposed in this paper can get the ranking of users' influence under different themes, and the correlation between the ranking of users' influence of different topics is low, and the high-impact users have higher information dissemination ability in many communities. The higher the ranking of users, the more evenly they spread across communities.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP393.09
本文编号：2459042
[Abstract]:With the development of Web2.0, online social network has gradually become the most popular communication platform in the Internet. Users in social networks can get to know each other and share and spread information. The mass of users in social networks and the information published by users have great business value and research value, which attracts more and more researchers to study social networks from different perspectives. Evaluating the influence of social network nodes is one of the hot topics in the research of social networks. By evaluating the influence of social network nodes, we can find high-impact people in social networks, which is of great significance to advertisement delivery, information control and user behavior analysis. In this paper, we mainly study the evaluation method of social network node influence, in order to find the high-impact person in the social network more effectively. On the basis of the analysis and research of the current methods of evaluating node influence, this paper proposes a new method of evaluating node influence of social network, and validates and analyzes the effect of evaluating node influence by experiments. Current research on node influence in social networks has paid little attention to the differences in influence under different topics and the ability of users to spread information across different communities. In this paper, a new method for evaluating the influence of topic-based nodes is proposed. The method takes into account the possibility of spreading information among users and the extent to which information may continue to spread after it is transmitted, and combines the relationships between users in social networks. The factors such as user's posting and forwarding behavior and the information content published by users are analyzed. The influence ranking of users under different topics is evaluated by the degree of affinity density and social circle difference among users. In this paper, the users and Weibo information in Sina Weibo are used as experimental data sets, and the results of user impact evaluation based on this method are analyzed. Experiments show that the method proposed in this paper can get the ranking of users' influence under different themes, and the correlation between the ranking of users' influence of different topics is low, and the high-impact users have higher information dissemination ability in many communities. The higher the ranking of users, the more evenly they spread across communities.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP393.09
【参考文献】
相关期刊论文 前1条
1 Zhao-yun DING;Yan JIA;Bin ZHOU;Yi HAN;Li HE;Jian-feng ZHANG;;Measuring the spreadability of users in microblogs[J];Journal of Zhejiang University-Science C(Computers & Electronics);2013年09期
，本文编号：2459042
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2459042.html
