Research And Application Of Rumor Recognition Method Based On Emotion Analysis

With the rise of social networking sites,microblog has become the most popular social networking information platform at home and abroad with the characteristics of high efficiency,strong communication power and diversified information.It brings convenience for people to obtain and share information.At the same time,there are also a lot of network rumors.Rumors have brought serious negative effects on the country,society and individuals.At present,the most common rumor identification method on the social network platform is manual identification,which not only consumes a lot of human,material and financial resources,but also has low efficiency.Therefore,many experts and researchers use traditional machine learning methods to identify rumors.Although they have achieved some results,they usually select shallow features such as user attributes and content features when selecting relevant features.Therefore,in order to improve the recognition of rumor information,this paper combines the traditional rumor recognition model with the emotional characteristics of comments,and proposes a rumor recognition method based on xlnet tgcg model.The specific research contents are as follows:(1)The XLNet pre training language model is used to obtain the word vector containing context information.At the same time,the advantages of its dual flow self attention mechanism are used to solve the problem of polysemy of a word,and the ability to express deep semantic information is improved.(2)The deep semantic features of microblog text obtained by Text CDN-GRU network and the emotional features of microblog comments obtained by CPSO-GRU are spliced and integrated to enhance the input characteristics to improve the performance of the rumor identification model.(3)Based on the XLNet-TGCG rumor identification model,this paper designs and implements a rumor detection system,which is mainly composed of data acquisition module,front and back-end interaction module and rumor identification module.It can return accurate rumor identification results for users when they submit rumor identification requests.After many comparative experiments,it shows that the xlnet tgcg model proposed in this paper is effective on Sina Weibo data set,and the value and accuracy of the model reach93.8% and 94.3% respectively,which shows that the model can be better applied to the rumor identification task.