Analysis of network rumor dissemination and control mechanisms on Chinese social network : Sina Weibo

The social network has become a major source of information. The openness and swiftness of the network bring instant accessing to both true and false information. Sometimes the false information such as network rumors can mislead people from obtaining the true information and even create chaos. The social network platforms and the governments in different countries are making effort on enacting their own mechanisms to strive on eliminating the rumors. The aim of this thesis is to investigate the spreading of rumors on the Chinese social networking website Sina Weibo. The censorship policy of the Chinese government and the methods used to control the spreading of rumors are discussed. The latest rumor control policy is a particular focus. It states that, if a post on Sina Weibo is tagged as a rumor and it is retweeted more than 500 times and/or viewed/clicked more than 5000 times, the originator of the post can be charged and sentenced to imprisonment for up to three years. The penalty also includes limiting access to Sina Weibo. Previous research into rumor dissemination on social networks is reviewed, including the definition and content of the rumors, as well as rumor spreading models, such as DK (Daley-Kendal) and its variant MK (Maki-Thompson), improved ones based on SIR (susceptible-infected-removed) from the theory of epidemics such as Weighted CSR, SIHR and SIRe. Also covered is the status of internet censorship and rumors in China and the development of Sina Weibo and its rumor refutation system. The latest rumor control policy is examined as a case study. A dataset was collected using a web crawler system which was technically supported by a cooperative research group with the keywords in Chinese of “rumors, being retweeted 500 times”. Based on the data analysis and visualization, characteristics such as discussion of the topic over time could be observed. A supplementary lexical text analysis is also covered. The results show the attitude of people towards the case topic, although the effect of the latest rumor control policy was not obvious due to the size of the dataset. No certain result is generated to indicate the connection between people’s rumor spreading willingness and the promulgation of the latest rumor control policy. This research addresses the following specific questions: How are rumors generated? What is the relationship between rumors and the rumor refutation system used by Sina Weibo? What is the effect of the latest rumor control policy on the dissemination of rumors and the control of public opinion? This area of investigation would benefit from further research into rumor control mechanisms, especially for Chinese social networks, using automatic rumor detection and refutation systems based on deep learning and sentiment analysis.