【作者】 杨路；

【导师】 何德全；

【作者基本信息】 北京交通大学 ， 信息网络与安全， 2010， 硕士

【摘要】 随着近年来BBS电子公告栏和SNS社区网络的风靡,越来越多的网民参与到各式各样的网络虚拟社区中,以虚拟人的身份在虚拟社区中与进行话题交流和兴趣分享。这种交流和分享把网络虚拟人紧密的联系起来,形成了庞大的复杂网络。在这个复杂网络中,虚拟人如何相互影响、兴趣主题如何传播等问题受到了广泛的关注,并衍生出观点动力学等一系列理论。本文以复杂网络理论为基础,借鉴观点动力学的内容,对网络虚拟社区中的兴趣传播规律进行了研究。本论文主要讨论分享类网络虚拟社区,以豆瓣网虚拟社区为实验数据来源,使用爬虫技术获取其网页原始数据,并使用预处理程序对原始数据进行过滤、分析。利用处理后的最终数据,本文对网络虚拟社区和其中的虚拟人成员分别进行了建模研究。其中对豆瓣网虚拟社区的拓扑研究使用了复杂网络和社会网络分析法,模型中以虚拟人成员为节点、虚拟人之间的好友关系为连线。同时讨论了此网络虚拟社区的相关特征参数,包括平均最短路径、聚类系数、度分布的仿真研究,最终验证了豆瓣网虚拟社区符合无标度网络特征。对虚拟人的建模则基于组织行为学,并参考了国内外关于网络虚拟人行为特征的相关研究,将网络虚拟人成员分为：享受者、奉献者、交际者三个类别。接下来本文对疾病传播模型SIR模型进行了改进,提出了适合网络虚拟社区的兴趣传播模型,通过程序仿真的方法模拟了兴趣主题在网络社区中的传播特性和各虚拟人节点的状态转换规律,并通过豆瓣网实际数据验证了此模型的合理性和实用性。更多还原

【Abstract】 In recent years, as BBC and SNS prevail in Internet community, more and more netizens have been joining various virtual communities online. They communicate and share stuffs of interest in these communities as virtual persons, and therefore weave a giant and complex network. The pattern that virtual people affect each other and disseminate interest drew wide attention and opinion-dynamics came into being. This paper probes into the law of dissemination of interest from opinion-dynamic perspective.This paper focus on the Internet community which is famous for its sharing and takes data from Douban.com as the subject for study. The writer collected original data by crawler and used preprocessor to select information for analyzing. The analysed data have been used in modeling to look into the Internet community and its members. Complex network and social network analytic approach is used in the modeling of Douban.com. In this model, virtual persons are dots and friendships between them are lines. This paper also includes discussion of parameters of related features in virtual community, such as simulation study on average shortest-paths, clustering coefficient and degree distribution. As a result, it proves that Douban.com fits the features of scale-free network. The virtual persons model is built under the theory of organizational behaviour, has consulted researches concerning the features of virtual human behaviour on the Internet and divides virtual community members into three categories:receiver, dedicator and raver.This paper continued to present an interest dissemination model, which is altered from disease dissemination model known as SIR. It simulates the spread of interest themes and the state changing pattern of virtual individuals by program stimulation. This model is proved reasonable and practical by factual data from Douban.com.更多还原