Deep Learning In A Social Network For The Application Of Text Classification

With the rapid development of the Internet industry and the popularity of computer hardware,especially the popularity of smart phones,the mobile Internet industry has developed rapidly in recent years.Because of the decline of mobile traffic tariffs,the coverage of WiFi and other hardware facilities,Internet users also increased rapidly.The range of Internet users'age basically covers the entire segment of the population,from primary school students to the olds.The emergence of e-commerce shopping platforms(Taobao.com,JD.com,and Amazon.com overseas),the social networking products(Weibo,WeChat)and foreign social networking platforms(Twitter and Facebook)also spread widely.As more and more active users continue to update their status,a large amount of text data is posted on these social platforms.The users post their own status,evaluating products or events,including chat records.It shows some important information.It's urgent to get the most valuable information from the massive amount of text data.After the deep study of the protrusive related theories in 2006,researchers made great progress in many fields,such as image and speech recognition.Subsequently,more and more researchers with the feature of self-leading try to apply deep learning in the text classification in order to solve the problem of the feasibility of understanding natural language in text classification.They using it to learn the text syntax and semantic features actively.What's more,they learn the extracted information deeply.The characteristics of the features not only make it easier to extract the artificial features,but also help to solve some problems including the strong subjectivity.The main work of this paper includes:establishing the neural network model,preprocessing the text,segment the word,extracting features,studying Algorithm and using Echarts to visualize the result of the text classification experiment,etc.This paper proposes to make full use of the different features of the text and combined with Neural Network Model of Convolution Neural Network and Recurrent Neural Network which purpose is to implementing the classification of text data.The subject of this article are texts of social networks.After extracting emotion,location,and Part of Speech features from the article.Firstly,we make the experiment with emotion,location,part of speech and other single features.Secondly,Experiment with the data which were divided into any two of the three features.Thirdrly,combined and tested the three features.Experiments show that in the deep learning neural network model,the more information the input feature vector combination has,the better the text classification effect is.After this,saving experimental results,and using ECharts technology to display experimental results through front-end JSP pages.Developed this Text Classification Visualization System by using ECharts is to make the analysis of experimental results more intuitively and efficiently.