微博意见领袖及其话题影响力的相关性分析研究与实现
发布时间：2018-11-05 20:21
【摘要】：作为信息化发展的重要产物,微博凭借其开放、便捷等特征,使大众在更广泛的领域进行交流成为可能,并在极短的时间内完成了由产生到普及,再到重要舆论平台的蜕变。而源于现实的社会事件,借微博平台酝酿发酵、被意见领袖传播散布、经各类观点碰撞交锋,往往会引起话题内容上的演变。在上述话题演变过程中,如何发现意见领袖并准确衡量他们对信息传播的作用和贡献?如何量化度量博文的传播影响?都是深入研究和分析话题及其脉络的先决条件。本文在研究了大量已有工作的基础上,从宏观、微观两条线索进行话题分析:首先,从宏观上研究“话题相关意见领袖与话题间存在怎样的时序关联?”、“哪些意见领袖影响话题发展?”等问题;在此基础上,从微观上研究“不同类型的博主主要发布何种内容特征的博文?”、“不同内容特征的微博会对话题造成何种影响?”等问题。研究中采取实验和实证相结合的方式,以现实中的微博话题为例进行了分析实验。相关工作和创新点如下:1)采集“南京打护士”话题有关微博数据,结合事实情况分析了数据用于研究的可行性;基于话题中有限的用户关系,采用一种通用简便的方法,找出了话题中存在的意见领袖;建立了“意见领袖在话题上的影响力”模型,和“话题热度”模型,为后续工作提供了支撑。2)研究并实现了基于动态时间弯曲(DTW)距离的博主影响力度量技术。在宏观方面,基于DTW距离,区分不同的时间窗口和时间粒度,对“意见领袖在话题上的影响力时序”与“话题热度时序”之间的距离进行了度量,发现距离较小的两者之间存在关联关系,发现部分意见领袖在话题中起到了导控和推手作用,在宏观上把握了话题发展脉络。3)研究并实现了基于影响饱和度的关键微博发现技术。在微观方面,分析了“转发量”和“影响饱和度”两种排序方式下微博的内容特征,基于用户转发行为和影响覆盖范围,研究了“影响饱和度”的时序特征,提出了用于单条微博评价的“影响饱和度”模型;在此基础上,基于特定话题,通过实验发现了“转发量”和“影响饱和度”两种排序方式下微博内容上的较大区别和博主类型上的较大区别,从微观上量化分析了其在话题中的不同作用。综上,本文主要通过现实中网络话题“南京打护士”等事件,利用DTW距离,研究了意见领袖和话题之间的相关性,从宏观上分析了话题脉络;进而引入了微博的“影响饱和度”概念,从“转发量”和“微博影响饱和度”两种视角,对话题作了微观解构,为现有的话题分析方法作了有益补充。
[Abstract]:As an important product of information development, Weibo, with its characteristics of openness and convenience, makes it possible for the masses to communicate in more extensive fields, and in a very short period of time has completed the transformation from production to popularization, and then to an important platform of public opinion. However, the social events originated from reality, brewing and fermenting by Weibo platform, are disseminated by opinion leaders, and often lead to the evolution of topic content after various views collide with each other. In the course of the evolution of the above topics, how can opinion leaders be identified and accurately measured for their role and contribution to the dissemination of information? How to quantify the impact of blog posts? Both are prerequisites for in-depth research and analysis of topics and their context. Based on the study of a large number of existing work, this paper makes a topic analysis from two clues, macro and micro: first of all, from the macro point of view, "what is the temporal relationship between the opinion leader and the topic?" "which opinion leaders influence the development of the topic?" On the basis of this, we study "what kinds of blog posts are published by different types of bloggers?" and "what kind of influence will Weibo with different content characteristics have on the topic?" And so on. Taking Weibo topic as an example, the research adopts the method of combination of experiment and demonstration. The related work and innovation points are as follows: 1) collecting the data about Weibo on the topic of "Nanjing beating Nurses", and analyzing the feasibility of the data used in the study according to the facts; Based on the limited user relationship in the topic, a general and simple method is adopted to find out the opinion leaders in the topic. The "influence of opinion leaders on topic" model and "topic heat" model are established, which provide support for further work. 2) the technology of measuring the influence of bloggers based on dynamic time bending (DTW) distance is studied and implemented. At the macro level, based on the DTW distance, different time windows and time granularity are distinguished, and the distance between "opinion leaders' influence on topic timing" and "topic heat time series" is measured. It was found that there was a correlation between the two, and that some opinion leaders played a guiding and pushing role in the topic. 3) the key technology of Weibo discovery based on saturation is studied and realized. In microcosmic aspect, the content characteristics of Weibo under two sort modes of "forwarding quantity" and "influence saturation" are analyzed. Based on user forwarding behavior and influence coverage, the temporal characteristics of "influence saturation" are studied. The "influence saturation" model for the evaluation of single Weibo is put forward. On this basis, on the basis of specific topics, we find the great differences in content and types of bloggers under the two sorting methods of "forwarding quantity" and "influence saturation" through experiments. From the micro-quantitative analysis of its different roles in the topic. In summary, this article mainly through the reality network topic "Nanjing hits the nurse" and so on event, uses the DTW distance, has studied the opinion leader and the topic correlation, has analyzed the topic vein from the macroscopic; By introducing the concept of "influence saturation" put forward by Weibo, this paper makes a microscopic deconstruction of the topic from the perspectives of "forwarding quantity" and "Weibo's influence on saturation", which makes a useful supplement to the existing methods of topic analysis.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092
，
本文编号：2313335
[Abstract]:As an important product of information development, Weibo, with its characteristics of openness and convenience, makes it possible for the masses to communicate in more extensive fields, and in a very short period of time has completed the transformation from production to popularization, and then to an important platform of public opinion. However, the social events originated from reality, brewing and fermenting by Weibo platform, are disseminated by opinion leaders, and often lead to the evolution of topic content after various views collide with each other. In the course of the evolution of the above topics, how can opinion leaders be identified and accurately measured for their role and contribution to the dissemination of information? How to quantify the impact of blog posts? Both are prerequisites for in-depth research and analysis of topics and their context. Based on the study of a large number of existing work, this paper makes a topic analysis from two clues, macro and micro: first of all, from the macro point of view, "what is the temporal relationship between the opinion leader and the topic?" "which opinion leaders influence the development of the topic?" On the basis of this, we study "what kinds of blog posts are published by different types of bloggers?" and "what kind of influence will Weibo with different content characteristics have on the topic?" And so on. Taking Weibo topic as an example, the research adopts the method of combination of experiment and demonstration. The related work and innovation points are as follows: 1) collecting the data about Weibo on the topic of "Nanjing beating Nurses", and analyzing the feasibility of the data used in the study according to the facts; Based on the limited user relationship in the topic, a general and simple method is adopted to find out the opinion leaders in the topic. The "influence of opinion leaders on topic" model and "topic heat" model are established, which provide support for further work. 2) the technology of measuring the influence of bloggers based on dynamic time bending (DTW) distance is studied and implemented. At the macro level, based on the DTW distance, different time windows and time granularity are distinguished, and the distance between "opinion leaders' influence on topic timing" and "topic heat time series" is measured. It was found that there was a correlation between the two, and that some opinion leaders played a guiding and pushing role in the topic. 3) the key technology of Weibo discovery based on saturation is studied and realized. In microcosmic aspect, the content characteristics of Weibo under two sort modes of "forwarding quantity" and "influence saturation" are analyzed. Based on user forwarding behavior and influence coverage, the temporal characteristics of "influence saturation" are studied. The "influence saturation" model for the evaluation of single Weibo is put forward. On this basis, on the basis of specific topics, we find the great differences in content and types of bloggers under the two sorting methods of "forwarding quantity" and "influence saturation" through experiments. From the micro-quantitative analysis of its different roles in the topic. In summary, this article mainly through the reality network topic "Nanjing hits the nurse" and so on event, uses the DTW distance, has studied the opinion leader and the topic correlation, has analyzed the topic vein from the macroscopic; By introducing the concept of "influence saturation" put forward by Weibo, this paper makes a microscopic deconstruction of the topic from the perspectives of "forwarding quantity" and "Weibo's influence on saturation", which makes a useful supplement to the existing methods of topic analysis.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092
，
本文编号：2313335
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2313335.html
