Font Size: a A A

Research On Social Stratification Behavior In Virtual Space

Posted on:2021-05-22Degree:MasterType:Thesis
Country:ChinaCandidate:X M YuanFull Text:PDF
GTID:2428330629488949Subject:Engineering
Abstract/Summary:PDF Full Text Request
With the rapid development of informatization and digitization,the Internet has become the most important virtual space relative to the real space.Traditional methods study social stratification in virtual space based on objective indicators such as the opportunity and ability of network information resources,not involving the specific behavior of users using network resources.It is generally acknowledged that social stratification also exists in virtual space,but there are different views on whether it continues the social stratification structure in real space.In response to the above problems,based on the big data of user behavior provided by China Internet Network Information Center,the work done is as follows.Firstly,we map the social stratification between virtual space and real space.The sample data is divided according to the social class of users in the real space,and the social stratification characteristics are studied from the aspects of online time and online content.Calculate the mean and variance of online time data to observe the stability of online behavior of users at different classes.Vectorize the software process clicked by the user,use the Word2 vec algorithm to extract word vectors that can represent the behavioral characteristics of users at all levels,and calculate the duration of the user's attention consumption in different types of software.The results show that the virtual space continues the social stratification structure in the real space.Secondly,we extracted the behavioral characteristics of users in different classes.Extract the click software data of each type of users in the dimensions of education,economy and age.Then build the vocabulary,get the one-hot vector of each software process and calculate the word vector representation of the process.Word vector representations of processes were calculated based on one-hot vectors.The skip-gram model in Word2 vec algorithm is used to train the word vectors that can represent various user behavior characteristics,and k-means algorithm is used to cluster the obtained features.The experimental results show that the higher class users can make better use of network resources for office and shopping,while the lower class users mainly spend their attention on leisure and entertainment software.Thirdly,we propose the W2V-BP model and use it to identify users' online behavior data in social stratification.The neural network model(W2V-BP)based onWord2 vec algorithm was trained by gradient descent optimization and several iterations of adjusting parameters.The model takes the user's online behavior characteristics as input and the user's hierarchical category as output,and the recognition accuracy rate is90.22%.The research results show that there are behavioral characteristics can distinguish human social stratification in the virtual space.Finally,compared with the LSTM model and SVM model,found that the recognition rate of the W2V-BP model in this paper has increased by 5.54% and 3.06% respectively.
Keywords/Search Tags:Big Data on User Behavior, Virtual Space, Social Stratification, Focus of Attention
PDF Full Text Request
Related items