Design And Implementation Of A Visual System For Text Emotion Analysis

The rapid development of information technology and the solid foundation of network infrastructure construction have enabled more and more people to participate in comments on the Internet.The rapid development of the internet not only brings new things to people,but also brings rich perspectives and emotional confluence.People are increasingly willing to express their emotions online.It is necessary to use sentiment analysis and visualization techniques to analyze these viewpoints.The new progress in visual analysis is operational visualization.This article visualizes the training process of deep learning models and achieves a method that can intuitively experience deep learning training.The main research content of this article is as follows:(1)This study focuses on the development of sentiment analysis technology and selects suitable sentiment analysis methods for Weibo text by consulting relevant literature.A BERT-CNN-LSTM-ATT sentiment analysis algorithm model is designed,which combines the local text semantic extraction ability of CNN and the context text semantic extraction ability of LSTM.BERT is used as a pretrained model for dynamic text embedding,Solved the problem of insufficient understanding of text semantics due to the traditional Word2 Vec method of static word embedding.And the use of attention mechanism in the model enhances the weight of feature allocation and improves classification performance.(2)This study designs and develops a visualization system for emotional analysis.This system not only visualizes the process of emotional analysis,but also uses visualization technology to display the results of emotional analysis and visualize the training process of neural networks.It allows users to have an intuitive feeling of network status,set network parameters,and modify the model in real time.The system can also extract keywords from input text and display them using methods such as word clouds.