摘要
为了理解舆论的时空演变斑图并揭示其形成机理,本文运用统计物理学方法,通过对在线评论数据进行统计分析,定量地研究了在线热点事件关注度(在线评论数)的时空演变规律.实证表明,虽然事件关注度在不同地区的分布存在极大的异质性,即遵循双段幂律分布;但是不同地区内事件的关注程度在时间演变过程中却表现岀明显的一致性,其不同时间内的Zipf分布变化很小.关联性分析显示地区关注度受到该地区经济的显著影响,而不同地区关注度演变行为的一致性来源于地区之间的强关联性.另一方面,舆论引导将显著影响事件的关注度,导致传播速度(单位时间内评论数的增量)急剧增加.通过计算不同地区传播速度的信息熵,我们发现评论的地区分布在大部分时间内都具有一致性,而舆论引导有助于保持这种一致性.地区传播速度之间的关联性分析表明在整个事件中经济较发达地区的舆论变化更趋于一致,暗示这些地区对于舆论引导的响应更快,因此加强发达地区的舆论引导有利于控制舆论的整体传播.
To understand the spatiotemporal evolution of online public opinion and reveal its lbrmation mechanism, we investigate tl^e oata from several popular online comments by means of statistical physics. Although the empirical results show that the heterogeneity of concerns exists in different areas, which follows a double power law, an obvious consistency of such concerns occurs during the evolution of online public opinion. Through correlation analysis, we reveal that the regional population and economy may have a significant influence on the concern about the event, and find that the consistency of concerns in different areas derives from the strong correlations among regions. On the other hand, the public opinion guide can significantly affect the concern about the event, and lead to the rapid increase of propagation velocity. By calculating the information entropy of propagation velocity, we find that the geographical distribution of online comments is relatively stable in most time, and the public opinion guide may help to maintain this consistency. Furthermore, the correlation analysis shows that the more developed areas tend to be more synchronized, which suggests the responses of these areas to the public opinion guide may be faster. Therefore, enhancing the guide of public opinion in developed areas can help our government to control the spread of the online public opinion.
出处
《物理学报》
SCIE
EI
CAS
CSCD
北大核心
2012年第9期526-532,共7页
Acta Physica Sinica
基金
国家自然科学基金重大研究计划(批准号:90924011
91024026)
国家自然科学基金青年科学基金(批准号:11105025)和资助的课题~~
关键词
在线舆论
关注度
ZIPF分布
信息熵
online public opinion, attention, Zipf distribution, information entropy
