基于话题的网络舆情演化机制研究

发布时间：2018-05-05 20:01

本文选题：网络舆情 + 话题提取　；参考：《南京邮电大学》2015年硕士论文

【摘要】：互联网信息的开放式、交互式传播使其传播范围更广、传播速度更快,一些热点问题、焦点事件或敏感话题一经网络传播,其影响力会成倍放大,甚至会在瞬间将政府、企业、组织或者个人推向风口浪尖,网络舆情的监控和管理一直受到多方面学者的关注和重视,目前也取得一些成果,本文从话题的角度来探究网络舆情演化规律,将网络中混杂的信息进行合理的整理和组织,旨在从更细的粒度上来研究网络舆情的话题演化以及传播规律,为网络舆情的监控和管理提供一定的理论基础和方法指导,本文的研究内容主要分为以下几个方面:(1)网络舆情话题的提取。基于PLSA坚实的统计学基础,提出了网络舆情话题提取方法,通过实际案例分析,成功抽取到网络舆情话题,利用其统计学特性准确描述了话题、文本以及词之间的概率语义关系,并通过特征词及特征词对话题的贡献率定量定义了话题。(2)网络舆情话题的演化。首先,构建了网络舆情话题的静态演化模型,通过计算定义话题的特征词之间的关联度构造了关联矩阵,并利用社会网络分析工具绘制了网络舆情话题静态演化路径,研究发现,字面上没有关联的话题之间也会存在某种联系,而这种联系与特征词的共现概率有关;然后,构建了网络舆情话题的动态演化模型,将网络舆情话题系统看作一个动态网络,分析了动态网络的微观结构,并通过实证分析研究了网络舆情话题的动态演化特性:话题演化路径、热点话题演化路径以及焦点话题演化路径,从而更深入得揭示了网络舆情话题的演化机理。(3)网络舆情话题的传播。首先,根据经典传播模型SIR分析了话题的传播机制,并通过仿真实验,分析了话题传播的影响因素(用户行为和话题内容)对话题传播和扩散的作用机制,在此基础上,引入了话题关联度,通过研究多话题环境下,话题关联度对话题转播概率影响的基础上,提出了多话题传播模型,并研究了话题关联度、多话题发生时间间隔以及多话题初始传播节点对各个话题以及整个舆情传播的影响,仿真实验分析结果表明,本文所提的话题传播模型能够反映真实的话题演化规律,具有一定的实际意义。
[Abstract]:The open and interactive communication of Internet information makes it spread wider and faster. Once some hot issues, focal events or sensitive topics are spread on the Internet, their influence will be magnified, and even the government and enterprises will be immediately affected. Organizations or individuals push to the forefront, monitoring and management of network public opinion has been concerned and valued by many scholars, and some achievements have been made. This paper explores the evolution law of network public opinion from the point of view of topic. In order to study the topic evolution and dissemination law of network public opinion from a finer granularity, it can provide some theoretical basis and method guidance for monitoring and management of network public opinion. The research content of this paper is divided into the following several aspects: 1) extracting the topic of network public opinion. Based on the solid statistical foundation of PLSA, this paper puts forward the method of extracting the topic of network public opinion. Through the actual case analysis, the topic of network public opinion is extracted successfully, and the topic is accurately described by its statistical characteristics. The probabilistic semantic relationship between the text and the words, and the evolution of the topic of network public opinion is quantitatively defined by the contribution rate of the feature words and the feature words to the topic. First of all, the static evolution model of network public opinion topic is constructed, the correlation matrix is constructed by calculating the correlation degree between the characteristic words that define the topic, and the static evolution path of network public opinion topic is drawn by using social network analysis tool. The study found that there is also some connection between topics that are not literally related, and this connection is related to the co-occurrence probability of feature words. Then, the dynamic evolution model of the topic of network public opinion is constructed. This paper regards the topic system of network public opinion as a dynamic network, analyzes the micro structure of the dynamic network, and studies the dynamic evolution characteristics of the topic of network public opinion through empirical analysis: the path of topic evolution, The hot topic evolves path and the focal topic evolvement path, thus has revealed the network public opinion topic evolution mechanism. 3) the network public opinion topic dissemination. First of all, according to the classical communication model SIR, this paper analyzes the mechanism of topic propagation, and through simulation experiments, analyzes the influence factors (user behavior and topic content) on the mechanism of topic propagation and diffusion. Based on the study of the influence of topic correlation on the probability of topic transmission in multi-topic environment, this paper puts forward a multi-topic communication model, and studies the degree of topic correlation. The influence of the time interval of multi-topic occurrence and the initial transmission node of multi-topic on the spread of each topic and the whole public opinion. The simulation results show that the topic propagation model proposed in this paper can reflect the real topic evolution law. Has certain practical significance.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：D035;G206

【参考文献】