The Depth Neural Network For Microblog Short Text Sentiment Analysis Study

In recent years,with the gradually mature of social networks and the rapid development of mobile technology,Microblog as a form of the major network transmission media,get the favour of more and more people.Users to express ideas spread through on weibo,expressing personal feelings at the same time,but produced a large number of personal subjective emotion characteristic information,the information contained in the emotional characteristics of different trends,which can have a significant impact on the spread of network public opinion.Based on the background of deep learning,this paper on the Internet microblogging essay of the sentiment analysis of these data is studied.The concrete research content is as follows:(1)For Chinese Microblog essay emotion tendentiousness in this judgment,this paper proposes a convolutional neural network model based on the depth of emotion classification method.This method will first the term vectors of training as an original feature vector,and then the characteristic vector into Convolutional Neural Networks model is further feature extraction,training the convolution of the Neural network classification model,according to the classification model essay this emotion classification of Internet.Experiment compares the SVM algorithm based on traditional machine learning and deep learning of CNNs model of the random vector method and the method proposed in this paper,finally through the experiment result proves that this method can effectively classify emotions.(2)According to the theory of weibo short text evaluation objects extraction problem,this paper proposes a bidirectional long short term memory neural network model on the emotional elements extraction method.Through experiment contrast the traditional machine learning model and Recurrent Neural Networks,when Long Short Term Memory model and Bidirectional Long Short Term Memory model found that three deep learning model based on depth study of the Bidirectional Long Short Term Memory can handle evaluation objects extraction tasks get the best effect.