Determine The Authenticity Of The User Identity On Social Network

With the rapid development of Internet technology,social network has gradually played an important role in our daily life.However,there are a large number of special users such as spammer,machine account and advertising board,and their valueless information.which is deviated from the original intention and seriously affects the healthy development of social network.User identity authenticity detection in social network helps to discover real and valuable information.Based on existing research results,this article studied the authenticity of user identity in social networks and proposed a hierarchical user identity verification method based on deep learning and maximum entropy model,and then experiments is carried out to prove the feasibility and effectiveness of our proposed method.The major works of this article are listed as follows:Design of emotional analysis model of microblog text based on deep learning.The self-learning ability of deep learning model is used fully to reduce the complexity of the feature representation in textual emotional problems.For the colloquial and symbolic expression style of microblog short text,a processing module including semantic translation,substitution and polarity transfer.We design a model based on Bi-LSTM to deal correlation context and correction result.The experiment results demonstrate the accuracy in multi-classification problems increased by more than 5%.Construction of user identity authenticity evaluation model based on maximum entropy model is exploited fully to avoid the shortcomings of the feature independence hypothesis.Moreover,on the basis of our proposed feature classification criteria,the authenticity of user identity is assessed with user emotional and behavioral characteristics.Experiment results show that the greater the difficulty of data acquisition,the more obvious the advantages of the model.After all,the model can automatically adapt to the changing expression habit in microblog text by means of expanding training corpus to address the problem of model failure because of the development of internet user,which has good sustainability and practical value.