The Calculation Method Of Text Emotion Intensity Research Based On Continuous Dimension

As a popular research direction in the field of natural language processing,emotional computing has become a hot research topic for the calculation of emotional information contained in many texts on the network.Most of the previous emotional analysis research is mainly discrete class type research,prior to the emotion is divided into different categories,this analysis method there are many shortcomings.In recent years,there is also a method of emotion analysis based on continuous dimension,which converts the emotional information of emotional words into two successive values of valence(the positive and negative degree of emotion)and Arousal(the degree of calm and excitement),And then through the text contained in the emotional words to be calculated as the text of the emotional intensity value.This paper based on the continuous dimension of the emotional analysis method,At the level of the phrase text,there are modifiers around the affective word dictionary-based emotional analysis,we use the support vector regression(SVR)method and rules to complete the continuous dimension of the phrase Text emotion intensity prediction.Firstly,the Word2 vec is used as the model input,and the support vector regression(SVR)method is used to complete the prediction of the emotional word.Then,the training data are used to extract the different modification rules,finally the whole phrase emotion intensity prediction is completed.The experimental results show that the method can effectively predict the emotional intensity of Chinese and English phrases.In the emotional analysis of sentence text,A model(W-CNN-LSTM)based on linear weighting that combines a convolutional neural network and short-term memory(LSTM)is proposed to solve the shortcomings of discrete category type emotion analysis which can not reflect the need of pre-definition of emotional intensity and category and the fact that sentence emotion analysis can not accurately capture emotional information.The model uses CNN to extract the local emotionalinformation of the sentence and the logical relation in the LSTM capture sentence,and completes the sentence text emotion intensity prediction by linear weighting.The results show that the method can effectively capture the local emotional information and the logical relationship,and can accurately complete the sentence emotion emotion prediction.