Analysis Of The Textual Emotional Orientation Of Web Reviews

With the rapid development of the Internet,a large number of Web comments are generated on the network,which brings new opportunities and challenges to natural language processing.It is of great commercial and social value to dig deeply into the emotional tendency contained in the massive text comment information.However,it is difficult to identify and summarize the emotional tendency hidden in the abundant opinion resources by the text emotion classification technology based on the single deep learning model.Therefore,this paper mainly conducts research on Web text emotion analysis based on deep learning model,and the relevant work contents are as follows:(1)This paper discusses the research background and significance of text emotion classification,expounds the research status of emotion analysis and deep learning at home and abroad respectively,and summarizes the relevant theories of deep learning,including artificial neurons,activation function,neural network,loss function,gradient descent and back propagation.(2)The convolutional neural network(CNN)and cyclic neural network(GRU)are integrated,and the attention mechanism is introduced on the basis of neural network,which is reflected in two aspects: 1.The original CNN model is modified to capture the sequence features of the word vectors with CNN,and then the captured feature information is integrated into the word vectors,so that the model can selectively focus on important features;2.2.This paper adopts the idea of encoder-decoder of attention mechanism to construct GRU network,and USES the last hidden state of the Encoder as the input of Decoder in decoding to avoid unnecessary interference.The core of the attention mechanism is to selectively focus on the high-value information among a large number of miscellaneous information.The introduction of the attention mechanism in the text emotion classification task can quickly learn the text characteristics of the sentence and capture the internal structure of the sentence.(3)From the methods based on the emotions of the Dictionary,an emotional vector guide neural network capture text sequence information,further increase the reliability of the model of emotion classification,is proposed in this paper.Reference ntusd,National Taiwan University,National Taiwan University Sentimental Dictionary)simplified Chinese emotional Dictionary positive word Dictionary,negative words in the Dictionary,and construct the commonly used negative word Dictionary,turn word Dictionary and word Dictionary,a total of 5 kinds of Dictionary.The experiment proves that the method of constructing emotion vector auxiliary classification proposed in this paper improves the accuracy of the model.