以新浪微博为数据来源,选用含有“谣言”的微博总数代理微博用户对谣言的关注度,并识别其中的原创微博数量、认证微博数量、图片微博数量、视频微博数量和短链微博数量,进而对各变量之间的关系展开实证研究。结果发现:第二天的微博用户谣言关注度与前一天的原创微博数量、图片微博数量和短链微博数量显著相关;原创微博数量、认证微博数量、图片微博数量和视频微博数量均是微博用户谣言关注度的格兰杰原因;微博用户谣言关注度波动性大、预测困难,对自身冲击的响应有一周左右的惯性,其变化的预测方差主要由自身变化来解释。
With Sina Weibo as the primary data source, this study used the total number of micro-blog posts containing the keyword "ru- mor" as a representative of attention on rumors in social media users, and then counted the number of original micro-blog posts, micro- biog posts from authenticated users, posts with pictures, posts with videos and posts with links to examine the relationships between these variables and rumor attention. The results showed that rumor attention in social media users on the next day is correlated significantly with the number of original micro-blog posts, posts with pictures, and posts with links on this day. The number of original micro-blog posts, posts from authenticated users, posts with pictures and posts with videos are the Granger reasons of rumor attention in social media users. Rumor attention in social media users fluctuates wildly and is difficult to predict, and its impulse response to itself has a week or so inertia. The prediction variance of rumor attention is to a large extent explained by self-changes