基于链接分析的微博用户可信度研究

发布时间：2018-07-05 08:21

本文选题：微博 + 用户可信度　；参考：《兰州交通大学》2015年硕士论文

【摘要】：微博作为一种新型的社交网络平台正受到越来越多人的青睐,使用微博平台分享和获取信息已经成为人们交流的重要方式。由于微博的自发性、低约束等特点,以及缺乏对用户有效的管理与制约手段,导致在微博用户数量快速增长与微博信息飞速传播过程中,微博虚假信息及谣言泛滥。因此,微博用户可信度问题逐渐引起人们的关注,成为一个亟待解决的问题。微博信息属于互联网信息的一部分,其短文本格式和复杂的社交网络特性给微博用户及其信息的可信度研究带来了巨大挑战。目前,微博用户可信度研究方面取得了一些初步成果,但存在着评价方法单一,元素特征考虑简单,算法复杂度高等不足。本文基于现有微博可信度研究成果,以新浪微博为研究对象,运用社会网络平衡理论知识对微博用户间信任传递关系进行分析。并从链接分析角度出发,基于PageRank算法思想,结合用户的特征属性,研究了一种评估微博用户可信度的模型及算法,针对同一微博领域用户的可信度排序进行相应的实验及分析。论文主要研究内容如下:首先,论文分析了微博的结构特点,研究了新的交流模式对用户可信度行为造成的影响,综述了可信度研究相关理论与主要技术,探讨了微博信息可信度研究内容,对比分析了当前微博可信度研究模型和表示方法。其次,论文基于微博信息传播机制及社会网络平衡理论,分析微博用户之间的信任传递关系,发现稳定微博网络结构中可信度相近用户之间更易建立关联。基于PageRank思想,将微博用户可信度排序问题转化为群体用户的信任支持问题,而用户间的链接即是用户因需求而产生的信任支持行为,进而从链接分析角度提出了微博用户可信度评估模型。最后,论文分析微博用户的特征属性及相关行为关系,结合本文模型设计了相应的可信度排序算法。算法利用非平均权值的思想,更加强调用户粉丝质量的重要性,能够有效降低用户可信度得分排序结果中僵尸粉的影响。基于网络爬虫获得的微博群体用户数据的实验结果表明了算法的有效性。
[Abstract]:As a new type of social network platform, Weibo is being favored by more and more people. Using Weibo platform to share and obtain information has become an important way for people to communicate. Due to the characteristics of Weibo, such as spontaneity, low constraint and lack of effective management and restriction measures to users, Weibo false information and rumors are rampant in the process of rapid growth of Weibo users and rapid spread of Weibo information. Therefore, the issue of Weibo user credibility has gradually attracted people's attention and become a problem to be solved. Weibo information is a part of Internet information. Its short text format and complex social network feature bring great challenges to the credibility of Weibo users and their information. At present, some preliminary achievements have been made in the research of Weibo user reliability, but there is a single evaluation method, simple consideration of element features and high complexity of the algorithm. Based on the existing research results of Weibo credibility and taking Sina Weibo as the research object, this paper analyzes the trust transfer relationship between Weibo users using the theory of social network balance. From the point of view of link analysis, based on the idea of PageRank algorithm and the characteristic attributes of users, this paper studies a model and algorithm to evaluate the reliability of Weibo users, and carries out corresponding experiments and analyses for the ranking of users' credibility in the same Weibo domain. The main contents of this paper are as follows: firstly, this paper analyzes the structural characteristics of Weibo, studies the influence of new communication mode on user's credibility behavior, and summarizes the relevant theories and main technologies of credibility research. This paper discusses the research content of Weibo information credibility, and compares and analyzes the current research model and representation method of Weibo credibility. Secondly, based on Weibo information transmission mechanism and social network balance theory, this paper analyzes the trust transfer relationship between Weibo users, and finds that it is easier to establish the correlation between users with similar credibility in stabilizing Weibo network structure. Based on the idea of PageRank, the problem of Weibo user credibility ranking is transformed into the trust support problem of group users, and the link between users is the trust support behavior caused by users' requirements. Then, from the point of view of link analysis, the Weibo user reliability evaluation model is proposed. Finally, this paper analyzes the characteristic attributes and related behaviors of Weibo users, and designs the corresponding reliability ranking algorithm combined with this model. By using the idea of non-average weights, the algorithm emphasizes the importance of the quality of users' fans and can effectively reduce the influence of zombie powder in the ranking results of users' credibility scores. The experimental results of Weibo user data based on crawler show the effectiveness of the algorithm.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP393.092

【参考文献】