Research On Emotion Recognition Of Web Public Sentiment Based On The OCC Model And Deep Learning

The research on the web public sentiment has attracted serious concern in the big data era,and the emotion classification is one of the key topics in Natural Language Processing.Lacking of semantic understanding and relies excessively on the emotion dictionaries are the long-standing problems of the emotion classification.The OCC model defines a new way to establish emotion rules to complement those two deficiencies.The present paper based this exact rules to label the emotion of each micro-blog text as the dataset,and build up the emotion recognition model with the help of Convolutional Neural Network(CNN).The establishment of the emotion classification model based on the OCC model and the CNN following three steps:first,three types of emergencies were selected as a case study,each text are divided into three categories(negative,positive,neural)based on emotion rules development upon the OCC model.The accuracy of the rules is comparable to that of trained on human-labeled dataset.Secondly,text after labeling is represented in a word vector way by using Word Embedding method,and then the CNN is used as classifier to automating label emotion of each text Finally,the scientific nature of the OCC emotion rule tagging system,Word2Vec text representation method and the CNN emotion classification algorithmis is verified through a comparative experiment,and their impact on the emotion recognition model of web public sentiment is also studied.One of the contribution is that the present study builds the emotion recognition model from the perspective of emotion cognition and deep learning.The innovation of the methods of online users' emotion classification is achieved by adopting the OCC emotion cognitive mechanism to design emotion rules,and the micro-blog text can be classified into three emotion classes by using this normative rules.The technological innovation of automatic emotion identification is achieved in the field of emotion analysis of web public sentiment by applying the CNN method to automatic identify emotion state of online users' in an emergency.