基于微博的城市突发事件扩散特征研究

发布时间：2018-05-13 10:44

本文选题：微博 + 突发事件　；参考：《武汉大学》2017年硕士论文

【摘要】：微博凭借其庞大的用户群体,成功参与引爆一系列社会事件,比如地沟油事件,西南大旱事件等,这些事件均通过微博进入公众视线,并在微博推动下向前推进。我国正处在突发公共事件的高发期,而且在未来一段时间内,我国都将面临突发公共事件所带来的严峻考验。因此,通过微博数据探究城市突发事件的扩散特征,可以为政府应对突发事件提供参考建议,及时向社会发布信息,引导舆论走向。文章基于微博开放API,获取城市突发事件的数据,经过一系列数据预处理,将微博的文本信息,时间信息和地理信息进行扩散特征研究。利用ICTCLAS汉语分词系统,对微博的文本信息进行分词,通过词频分析法探究事件发生的完整过程。对时间信息进行每天每小时数据统计,研究用户使用微博的行为特征。利用重尾理论,分析微博的地理信息,发现突发事件发生后的地理扩散规律。重尾理论是一种病态的概率分布模型,尽管应用领域广泛,国内还没有把它应用到地理空间分析。然后与热点分析的结果进行比较,发现可行性。通过回归分析确定信息扩散与事件发生地点是否有关联。文章以"杭州公交燃烧案"为例,进行城市突发事件的扩散特征研究。将文本中排名前十的名词、动词、形容词进行排序,可以清楚地看出事件的走向,从怀疑是人为事件到嫌疑人身份的确认,只用了四天时间。微博用户的活动事件具有一定的规律性,总体上呈现幂律分布,每天有固定的活跃时间段。对比热点分析的结果,重尾理论可用于地理空间分析。在突发事件发生地点的周围,公众的关注程度受距离的影响,但大城市对事件的关注不太可能受到距离的影响。综上,充分探究城市突发事件发生后的微博数据,从新的角度分析其位置信息,是顺应快速发展和政务微博的需要。
[Abstract]:With its huge user base, Weibo has been involved in detonating a series of social events, such as the gutter oil incident, the southwest drought and so on, all of which have been brought into the public eye through Weibo and pushed forward by Weibo. China is in a period of high incidence of public emergencies, and in the future, our country will be faced with the severe test brought by public emergencies. Therefore, exploring the diffusion characteristics of urban emergencies through Weibo data can provide reference suggestions for the government to deal with emergencies, release information to the society in time, and guide public opinion. In this paper, the data of urban emergencies are obtained based on Weibo. After a series of data preprocessing, the diffusion characteristics of text information, time information and geographic information of Weibo are studied. Using the ICTCLAS Chinese word segmentation system, the text information of Weibo is partitioned, and the complete process of the event is explored by word frequency analysis. The daily hourly data of time information are analyzed to study the behavior characteristics of users using Weibo. Based on the theory of heavy tail, the geographic information of Weibo is analyzed, and the law of geographic diffusion after unexpected events is found. Heavy-tailed theory is a ill-conditioned probability distribution model. Although it is widely used in many fields, it has not been applied to geospatial analysis in China. Then compared with the results of hot spot analysis, the feasibility was found. Regression analysis is used to determine whether the information diffusion is related to the location of the event. Taking the case of Hangzhou bus Burning as an example, this paper studies the diffusion characteristics of urban emergencies. By sorting the top ten nouns, verbs and adjectives in the text, we can clearly see the trend of the events, from suspicion of human events to confirmation of suspects' identity, in only four days. The Weibo user's active events have certain regularity, which is power law distribution in general, and has a fixed active time period every day. Compared with the results of hot spot analysis, the heavy-tailed theory can be used in geospatial analysis. The attention of the public is affected by distance, but the attention of big cities is not likely to be affected by distance. In summary, it is necessary to fully explore the Weibo data of urban emergencies and analyze the location information from a new perspective. It is necessary to adapt to the rapid development and the need of government Weibo.
【学位授予单位】：武汉大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：D63

【相似文献】