Microblog Rumor Detection Based On User Behavior Characteristics

Living in an era when the Internet is developed,people generally use the Internet to communicate.The network communication tool represented by Sina Weibo has a huge amount of information,which brings great convenience to the fermentation of rumors.The emergence of Internet rumors not only brings panic to people,but even causes irreparable harm to people.Using Sina Microblog as the research platform,focusing on user behavior characteristics,user information characteristics,and text content characteristics,with the goal of improving the accuracy of feature-based rumor detection,integrating user behavior characteristics,user information characteristics,and text content into the model Features,build the BERT-DPCNN model,and the ERNIE-DPCNN model for network rumor detection research.Two main tasks have been done in the article:The first one is to propose a microblog rumor detection method based on user information characteristics and microblog text characteristics.Most researchers use a single text content as the main target of rumor detection,ignoring the role of other features.And when using the bag-of-words model to process the long text content of Microblog,part of the data will be forgotten.In response to these two problems,this article proposes to combine user information characteristics to build a BERT-DPCNN model for rumor detection.The second is to propose a microblog rumor detection method based on user behavior characteristics.Although the BERT-DPCNN model improves the accuracy of feature-based rumor detection,the user behavior characteristics have not been deeply mined.In order to further explore the help of Microblog user behavior characteristics for rumor detection,this article proposes to combine user behavior characteristics to construct an ERNIE-DPCNN model.The model uses user behavior characteristics such as users' likes,reposts,and comments as the main features,combined with the improvement of the deep learning model,to detect Microblog rumors events.A large number of experiments show that the rumor detection model proposed in this article can effectively improve the accuracy of rumor detection and has certain reference value.