Design And Implementation Of An Epidemic Sentiment Analysis System Based On Deep Learning

With the development of the Internet,the life pattern of human beings has changed dramatically.More people choose to express their ideas and opinions through the network and social platforms,share their daily life and interact with others.The number of users of Sina Weibo has reached hundreds of millions,in which the published texts,pictures,and other information present the different views of different people,and have great research and social value.The information in social platforms is easy to spread,and the content it represents may contain a lot of emotional information.Therefore,people's general ideas can be captured in time by observing and studying these emotional text data,which provides a positive effect on the healthy development and stability of society.At the beginning of 2020,the COVID epidemic spreads in China,which connected the people of the whole country together.Thus,the development of the epidemic attracted everyone's attention during that period.Sina Weibo has just become the best way for netizens to acquire,comment,and forward the latest news.People express their concerns and worries through comments and other methods.Therefore,it is necessary to analyze the sentiment of texts on the topic of epidemics on Weibo published by netizens.This paper proposes a new network model combining part-of-speech vectors to classify text content by applying the method of expanding the basic sentiment dictionary and constructing the sentiment dictionary under the proprietary domain based on deep learning respectively.The contents of the paper are as follows:(1)The construction of an epidemic-specific dictionary in the field of Sina Weibo.Through the expansion of the basic sentiment dictionary,this paper has collected a large number of Internet terms and sentiment words with characteristics of the Weibo domain to jointly construct a proprietary sentiment dictionary related to the Weibo epidemic field.Also,the classification of comments on Weibo focusing on the topic of the epidemic is more accurate according to the set scoring rules.(2)Because the traditional deep learning network model is relatively single and the input data is not sufficiently processed,this paper combines part-of-speech features on the input side,and uses Convolutional neural networks,Bi-directional long short-term memory,and self-attention mechanism as the framework to propose SA-Conv Bi LSTM to deals with sentiment analysis tasks.The effectiveness of the proposed model framework is proved by using evaluation indexes and comparative experiments.(3)The above two sentiment analysis methods were designed and implemented in python by utilizing the deep learning framework-Keras,and an epidemic sentiment analysis system was developed and implemented based on sentiment dictionary and deep learning.Through this system,it is convenient to upload data and download prediction results,conduct comparative experiments of various model frameworks,which have the functions of single-input and batch-input prediction and emotion retrieval.