Research On Internet Rumor Identification Based On Text Analysis

The Internet is profoundly affecting and changing the politics,economy,culture,and life of our society.The social media such as Weibo,WeChat,Twitter,etc.that came into being have brought convenience to people,and the rapidly spreading rumor information has become An obvious issue in the Internet age.The negative effects brought about by the generation and spread of rumors can easily stimulate social conflicts and affect the personal life and even the harmony and stability of the country and society.Therefore,how to identify and stop the Internet rumors in a timely and accurate manner before they spread widely.At present,Sina Weibo in China is one of the most active social networking platforms.Free and convenient information dissemination is a major feature of Weibo,but this feature also facilitates the breeding and spread of rumors.Therefore,this paper selects Sina Weibo as the data source to conduct research on the identification of online rumors.First of all,because the current research work on Internet rumors is relatively mature,but due to the differences in age,gender,or culture of users using Weibo platforms,different topics actually have different proportions in Internet rumors.Therefore,this article first analyzes network rumors based on text content and classifies network rumors according to themes.Based on the factors identified by predecessors(ie,word features,symbol features,emotional features,etc.),the characteristics of rumor recognition under each topic are analyzed.Secondly,this paper regards network rumor recognition as a binary classification problem,selects a specific rumor theme,combines the characteristics of network rumor recognition under the proposed specific topic,and uses the data mining software WEKA to previously construct a specific topic network The rumor recognition model is verified.The experimental results show that,based on the text content,the use of topic-based network rumor recognition features has a better effect on specific network rumor recognition,that is,the network rumor recognition features constructed based on specific topics are effective.