Design And Implementation Of Microblog Rumor Recognition System Based On Deep Learning

In recent years,with the rapid development of Internet technology,the amount of information on social platform has increased dramatically,bringing great changes to people's lives.At the same time,along with a lot of rumors.Once the rumors are spread,the harm caused is difficult to estimate,and its spreading speed and breadth are quite amazing.So how to efficiently identify rumors and reduce the hazards caused by it,undoubtedly become a problem that needs to be solved.In view of this,all walks of life and government departments have taken different measures to seek rumors through relevant technologies,and to maintain the fairness and normal order of society.As an important text classification task in natural language processing,the meaning of rumors is unquestionable,it is unlikely part of it.The current rumor recognition method is mainly based on manual judgment.This method is time consuming,and it has caused many researchers' attention,and people strive to identify rumors more efficiently through a variety of machine learning or depth learning models.In many social platforms,Weibo has become the mainstream public opinion communication medium due to its unique advantage,so this article selects Weibo as a platform for rumors.After research on a series of text classification methods,this paper proposes a rumor recognition method for the integration depth learning model,which proves its efficiency,thereby establishing a microblogging recognition system.(1)Based on the investigation of the existing rumor refuting platforms,this paper proposes a deep learning fusion model,which uses the Roberta model with good effect as the natural language pre training model to convert the microblog text into vector representation.Three text convolution kernels of different sizes are used to learn the features of microblog text,and the corresponding feature sequence is obtained after the maximum pool splicing operation of these features,and then input to Bi-LSTM to further learn the sequence features.Finally,the attention mechanism is added to calculate the attention distribution probability,so as to achieve the purpose of rumor recognition.The experimental results on two public microblog datasets show that compared with other methods,the proposed method has significantly improved the performance of rumor recognition,and can mine the deep features of microblog text.(2)After verifying the effectiveness of the depth learning fusion model,this paper is designed with the microblogging identification system based on this model,and the microblog information is discriminated with the well-trained deep learning fusion model,and the results are displayed to the user.For users to further analyze or use.