Study On The Influence Factors Of Weibo Retweeting In Food Safety Events

With the rapid development of economy and industry,people's pursuit of food has been raised to a green and healthy level from meeting daily needs.However,in the process of planting,transporting,processing and selling food,the participants' information asymmetry with each other resulted in frequent occurrence of food safety events.The issue of food safety is like a face-lifting terrorist activity.It must be taken seriously by the government departments.Weibo(without special explanation,means "Sina Weibo")has various modes of communication,including comments,attention,retweeting,etc.,and retweeting is the main way.For messages that are retweeted more,the propagation is explosive and nuclear fission.In the event of food safety,governments who are in the "micro-era" must timely control the situation and maintain the normal order of the network.However,the majority of government Weibo operators are unable to timely release Weibo with high influence because they do not understand the Weibo users' behaviors and the characteristics of Weibo content,resulting in the government's Weibo being in a "gland state".How to do will be the most effective?It is extremely urgent to study the influence factors of Weibo retweeting.Based on the definition of food safety events,the review of domestic and foreign literature,the article takes the "Haidilao event" with large social effect in 2017 as an example,to sort out the time points of event and analyze public opinions of the event based on Baidu index data.The baidu index data also discloses the event public opinion period and keywords,and collects Weibo data accordingly.After the data is cleaned,the emotion analysis is performed on the Weibo based on a dictionary method to understand the trend of Internet users' emotional changes.The Weibo emotions variables and other variables are used as input variables,and the retweeting variable is used as the output variable to construct a random forest model.Using the data of "Haidilao event" as the training set,and the data of "Eleme event",that were exposed at "the 315 party" in 2016 as the testing set to build models together.To extract important variables that affect the classification model.Due to random forest lose the ability to interpret variables,so that,it cannot explain whether an important variable is conducive to retweeting or inhibiting to retweeting,so the important variables of the model and the retweeting variables are analyzed in Crosstab to extract the important features that are conducive to Weibo retweeting.The results show that the Weibo users who are at the event area,Weibo level are above 40,gold V certification or official or personal certification,the number of Weibo on the scale of 10001-100000,the number of following on the order of 1001-10000,the number of fans is above 10001,the messages published can be easily retweeted.Weibo posts with the "[title]"symbol can be easily retweeted.And the above variables passes the Chi-square test,indicating that the difference between the amount of retweeting is affected by the above variables.