Research On Microblog User Emotional Portrat Model Based On Deep Learning And Its Application

The wide application of social media such as Weibo in the Web 2.0 era has had a great impact on people's daily life and production.The sentiment analysis of Weibo texts has also become a popular research direction at this stage.However,related research is still in its infancy,and there is no sentiment dictionary specifically for Weibo texts.The deep neural network model can realize the emotion classification of the text through a limited number of unsupervised autonomous training,avoiding the cumbersome manual labeling of corpus under massive data.Therefore,this paper proposes a Weibo user emotion portrait model based on deep learning technology,constructs a Weibo emotion dictionary and emoticon dictionary,and realizes the emotion classification of Weibo topic text with the help of a deep neural network model.First,integrate and optimize existing emotional dictionary resources to reduce the difficulty of dictionary construction.The iterative speed of internet buzzwords on Weibo is very fast.For this reason,a word vector learning model HIT based on deep neural network is introduced to obtain word vectors containing emotional information,combined with traditional classification dictionaries to build a Weibo sentiment dictionary;grab the 203 Weibo emoticons commonly used on the Internet have constructed a Weibo emoticon dictionary.Secondly,in order to better extract the emotional features of the Weibo text,this paper proposes a recurrent convolutional neural network model RCNN,which combines the advantages of the recurrent neural network RNN and the convolutional neural network CNN,and uses the convolution of the RNN recurrent layer and CNN The first-level and secondlevel feature extraction is performed on the layer and the pooling layer.At the same time,in order to solve the problem that CNN ignores the structure of the text sentence and cannot extract the sequence information,the strategy of segmented pooling is adopted;in order to improve the accuracy of the emotion classification of the model,in the double Based on the word topic model BTM and JST topic model,a two-word topic emotion joint model JSBTM is proposed to extract the topic features of the text,and use TF-IDF to calculate the feature weights.Finally,we integrated the previous two steps to complete the microblog user emotional portrait model based on deep learning,and carried out training.The classification effect of the model was compared experimentally using evaluation indicators such as accuracy,precision,recall,and F1 value.