Social Classes Sentiment Analysis In Social Media With Big Data

With the rapid development of Internet technology,mobile Internet has also been widely spread.Social media,as one of the main manifestations of the Internet,has become an important part of people’s lives.Its wide range of user participation makes the analysis of social media become a hot field.However,the current research on social media covers a wide range,and the real social users scatter in different social class.Different social class hold different views and demands.We believe that the research on social class based on social media can achieve a meticulous analysis of the objective social class and the views of different social class.And how to accurately and effectively analyze the social media in the social class and the attitudes of different social classes towards the certain product and event,are of great important reality meaning.It has positive effect on public opinion monitoring,marketing,rumor control and so on.It also plays an important role in promoting the national economic development,maintaining social stability and the overall social progress and so on.This paper aims to realize the social class emotion analysis for the social media big data background by designing the analysis method of the occupations,location,emotion and entity of the users in the social media big data.This paper studies the three key issues of social class emotional analysis in the context of large social media data,including the analysis of social class based on career,the analysis of social class based on user location and the analysis of social class emotional entity based on social media.The main research contents and innovations are summarized as follows:1.A social class classification method based on career analysis model is proposed.The method is based on the occupational social stratification model.And it can improve the accuracy of the classification of microblog users by means of constructing the professional lexicon and selecting the appropriate classifier to construct the combination feature extraction model.In this process,firstly,based on the theory of social stratification and the new social class theory proposed by Chinese President Xi Jinping,the model of social class based on career is designed and the combination prescribing test and TF-IDF algorithm are designed to establish the professional feature lexicon Mixed model.The professional thesaurus for each social class is established.By using the word lexical statistical algorithm,cosine similarity algorithm of space vector model,k-means algorithm based on Euclidean distance and SVM algorithm,comparing the accuracy rate,recall rate and F value of the result,the SVM algorithm is best suited for occupational classification.According to the SVM algorithm,by adjusting the input parameters c and g,the relationship between the parameters and the accuracy of the classification is obtained.The c and g with the highest accuracy are obtained,and the optimal classification algorithm applied in the microblogging user’s career analysis is obtained.According to the occupation based on the social class stratification model,the classification of the user is finally obtained.2.A location-based user social class analysis model is proposed.This model designs a social stratification model based on the economic status.It integrates the position estimation model RTP-LI of the landmark image semantic analysis,microblog text and social network analysis.Based on the multi-dimensional analysis of the user’s social media data,including the user social relations,blog text content,blog image and other data,a local dictionary suitable for analyzing short text is established.And then,by introducing the accurate Sogou dictionary,a comprehensive thesaurus based on user’s location when he sends text gets constituted.Through the result of the landmark image semantic recognition,the analysis of location is further amended,reducing the error caused by the text divergence.The divergence of the content of the whole text is solved by the method based on the combination of the user-generated content and the statistical analysis based on the user’s social network.Geographic information depends on a wide range of elements,this paper combines text-based,picture and social network,greatly enhance the accuracy of location inference.And ultimately user’s social class gets inferred based on the user’s geographical location.3.A mixed emotional Chinese entity recognition model is proposed.The model constructs the emotional analysis model and obtains the emotional tendency of microblog.This model designs the Chinese named entity recognition model based on the three-layer neural network to identify the entities behind the emotion,and then deduce the emotions of different social strata and the entity factors that form the emotion.In this process,we first design the emotion analysis model based on support vector machine,obtain the emotional tendency of microblogging,and carry on the emotion sense sign of the characteristic words in the blog post.And then design the algorithm model based on the three-layer neural network,preprocess the corpus of The people’s daily to generate the word vector dictionary and training matrix of the Chinese named entity,and train the three-layer neural network model.Finally,the trained model is used to identify and obtain the Chinese named entity with the emotional orientation.