Research And System Implementation Of COVID-19 Rumor Recognition Based On Machine Learning

In recent years,the rapid development of social media has become the main platform for people to obtain information and exchange opinions,especially Twitter,Weibo and other platforms,while providing convenience for people to obtain information,it also breeds a lot of rumors.In the social media environment,rumors are constantly breeding and spreading extensively,which has become a severe social problem and often leads to serious consequences.Especially since the outbreak of the new crown epidemic,rumors related to the epidemic are endless,causing panic and even social confusion.For the tens of thousands of information on social platforms every day,the efficiency of filtering and screening rumors only by manually verifying the rumor is very low.Therefore,how to identify rumors as soon as possible is an important measure to take effective measures to suppress rumor dissemination in a timely manner.premise.This article uses Weibo data as the starting point.Based on machine learning methods,we analyze and build rumor recognition models based on machine learning.At the same time,the construction model design and realize the visualization system of the new crown epidemic rumors.The specific research content is as follows:(1)Feature extraction and analysis of COVID-19 rumors: Based on microblog data,5 features were extracted from content information,12 features were extracted from user information,and 7 features were extracted from spreading information.Furthermore,statistical analysis and machine learning methods are adopted to analyze the transmission characteristics and user characteristics based on the commonness of rumors,and the unique text characteristics of epidemic rumors are analyzed.By analyzing the commonness characteristics of rumors,it is found that there are obvious differences between rumors and non-rumors in user characteristics and transmission characteristics.The emotional tendency analysis of epidemic data found that people expressed more negative emotions in epidemic rumors.(2)New crown epidemic rumor identification based on machine learning methods: The model-based model and model-based model-based models are constructed.Use the rich features extracted in the model-based model as the input of support vector machines and decision tree algorithms.In the epidemic data concentration,the support vector machine obtains 89.62%accuracy,and the decision tree obtains87.19%accurate accuracy of the accuracy of 87.19%.Rate;in the method model based on deep learning,the GCNS-Bert model based on the convolutional network is proposed.In the experiment,first using the nonepidemic data set to verify the effectiveness of the GCNS-Bert model,and obtained a 95.6%accuracy rate.The experimental results showed that the GCNS-BERT model has a good rumor recognition effect;then in the new crown epidemic data concentration data concentration Rumor recognition obtained a 92.5%accuracy rate,which was increased by 2.9%compared to the accuracy of the support vector machine.Experiments show that the GCNS-BERT model can effectively learn the hidden characteristics of the epidemic data and the hidden characteristics of the spread of the epidemic data and get better rumors.Identification effect.(3)Design and implementation of the new crown epidemic rumor recognition system: Based on the GCNS-Bert model design and realized the new crown epidemic rumor recognition visualization system,the system has data collection functions,user interaction functions,rumor recognition functions and data visualization functions.Essence The data collection function is mainly responsible for climbing the data of Weibo posts;user interaction functions mainly provide services to users;rumor recognition function to predict whether the Weibo post is a rumor based on the model;Display of prediction results.To sum up,this article conducts feature extraction and analyzes the common characteristics of rumors and unique features of the rumor for Weibo data.This model can improve the efficiency of epidemic rumors recognition.Therefore,based on this model,the new crown epidemic rumor recognition visualization system is achieved.