多媒体网络舆情危机监测指标体系构建研究

发布时间：2018-07-21 14:49

【摘要】：随着4G网络的发展及移动终端的普及,公众表达情绪、发表观点的方式由传统信息媒介转移到多媒体互联网平台上,促使多媒体网络舆情信息传播走进公众视野。公众通过APP软件、智能手机、各类移动终端便携、快速的参与到微博、微信、论坛等媒介承载的网络舆情信息中。与传统的报纸、电视等媒体相比较,多媒体网络舆情具有传播速度快、信息数量庞大、存储、管理、分析困难,舆情危机形成概率高,影响范围广,影响力强等特点。一旦形成舆情危机,损失不可估量。有关部门将多媒体网络舆情危机的及时监测与分析,作为引导、疏导舆情危机的先决条件。指标体系的构建正是危机监测的基础和依据。本文旨在构建多媒体网络舆情危机监测指标体系。首先,在研究多媒体网络舆情传播过程、分析影响舆情演进要素的基础上,本文提炼涵盖舆情产生发展的5个要素(舆情主体、舆情客体、舆情本体,舆情媒介、舆情环境)作为指标体系中的一级指标,同时在研究国内外文献及权威机构指标体系,建立了二级指标,并给出可量化的末级指标。整个指标体系客观性较强,具有广泛的代表性;其次,针对现有多媒体网络舆情信息量庞大的特点,本文利用Hadoop大数据处理平台,利用聚类,分类,关联规则等数据挖掘技术进行动态数据采集,快速处理图片、文本、视频等半结构化、结构化、非结构化的数据,为多媒体网络舆情危机的实时监测提供思路;再次,对指标进行筛选,采用相关性分析、主成分分析相结合的方法。先利用相关性分析方法,除去同一层指标集内相关系数较大的指标,确保剩余的指标体系能够表征初始指标集信息,降低初始指标集的冗余程度,之后再采用主成分分析法,通过线性变换,将剩余指标集进行降维处理,并借助累计方差贡献率思想验证最终的指标体系冗余度较低且信息表征能力依然很强;最后,采用基于遗传算法的BP神经网络确定筛选后各级指标的权重,利用德尔菲法与SPSS分析相结合方法设定初始控制参数及舆情危机等级,在降低人为主观因素影响评价结果的基础上,在宏观上映射出指标体系重要性,在微观上展示确定指标权值的客观性。本文以“韩国萨德”事件为实例模型,验证本文构建的多媒体网络舆情危机监测指标体系的合理性及必要性。
[Abstract]:With the development of 4G network and the popularity of mobile terminals, the way of public expression and expression of opinions is transferred from traditional information media to multimedia Internet platform, which promotes the dissemination of multimedia network public opinion information into the public view. Through app software, smart phone, various mobile terminals portable, the public quickly participate in Weibo, WeChat, forum and other media to carry network public opinion information. Compared with the traditional media such as newspapers and TV, multimedia network public opinion has the characteristics of rapid transmission, large amount of information, storage, management, analysis difficulties, high probability of public opinion crisis formation, wide range of influence, strong influence and so on. Once the public opinion crisis, the loss is incalculable. Relevant departments take the timely monitoring and analysis of multimedia network public opinion crisis as a prerequisite for guiding and unblocking public opinion crisis. The construction of index system is the foundation and basis of crisis monitoring. The purpose of this paper is to construct a multi-media network public opinion crisis monitoring index system. First of all, on the basis of studying the transmission process of multimedia network public opinion and analyzing the factors influencing the evolution of public opinion, this paper abstracts five elements (public opinion subject, public opinion object, public opinion ontology, public opinion medium) that cover the generation and development of public opinion. The environment of public opinion) is the first class index in the index system. At the same time, the index system of domestic and foreign literature and authoritative organizations is studied, the secondary index is established, and the final quantifiable index is given. The whole index system has strong objectivity and extensive representativeness. Secondly, in view of the huge amount of multimedia network public opinion information, this paper uses Hadoop big data processing platform, using clustering, classification, Association rules and other data mining technology for dynamic data acquisition, fast processing of images, text, video and other semi-structured, unstructured data, for multimedia network public opinion crisis real-time monitoring ideas; third, The method of combining correlation analysis and principal component analysis was used to screen the indicators. Firstly, the correlation analysis method is used to remove the index with large correlation coefficient in the same index set, to ensure that the remaining index system can represent the information of the initial index set and reduce the redundancy degree of the initial index set, then the principal component analysis method is used. Through the linear transformation, the residual index set is reduced, and the contribution rate of cumulative variance is used to verify that the redundancy of the final index system is low and the information representation ability is still very strong. BP neural network based on genetic algorithm is used to determine the weight of every index after screening, and the initial control parameters and public opinion crisis level are set by combining Delphi method and SPSS analysis. On the basis of reducing the impact of human subjective factors on the evaluation results, the importance of index system is mapped out in macro level, and the objectivity of determining index weights is shown in micro level. Taking the case model of "Korea sad" as an example, this paper verifies the rationality and necessity of the multi-media network public opinion crisis monitoring index system constructed in this paper.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：G206

【参考文献】