Research On Emotional Tendency Of Food Safety Information Based On Public Opinion System

With the rapid development of Internet technology,text data is exploding every day.This data contains dynamic information published by users on social media platforms such as Weibo,WeChat,forums,news content of major news media portals,The review information of products on various e-commerce websites,post emails communicated by users,etc.These text data seem complex,but in fact contain a lot of useful information,including public opinions and emotional tendencies.In recent years,research on text sentiment analysis has gradually increased.These studies can help companies understand the user's emotional attitude towards a certain product or service,and can also provide policy support for related units to formulate some policies in response to certain hot public opinion events and promptly guide them.Negative emotions,etc.This article mainly starts with food safety information that is closely related to people's lives.Taking "African Swine Fever" as an example,with the help of the original comment data of the public on the "African swine fever" incident in the public opinion system of China's Jilin Network,the machine learning method is used to analyze it.Perform emotional orientation analysis.The main research contents are as follows:(1)Overview of sentiment analysis theory.As an important research area of natural language processing,sentiment analysis needs to understand and learn related theoretical knowledge,including natural language processing methods,concepts and goals of sentiment analysis.In different application scenarios,what is needed to complete sentiment analysis The task,and the sentiment analysis can be divided into three levels according to the granularity of the text being processed.(2)Research on key technologies.The first task of sentiment analysis of Chinese texts is to preprocess the text,including Chinese word segmentation and stop words.Then,the text-represented model is used to represent the pre-processed text as an internal structure that can be recognized by the computer,so that the computer can use the text-represented model to perform numerical operations.Finally,there are many feature words after preprocessing,and the feature set will be very large.Feature selection is required.The purpose is to find the optimal feature subset,reduce computational complexity,reduce running time,and improve model accuracy.(3)Comparative analysis of classifier experiments based on machine learning.This paper uses a machine learning-based sentiment analysis method.In order to verify whether this method can deeply explore the impact of semantic features on sentiment analysis results,it innovatively uses three feature selections: bag-of-words model,TF-IDF,and Word2 vec based on neural network.And representation methods,combined with Naive Bayes,support vector machine,and logistic regression algorithm,using the "3×3" method to design 9 sets of combination methods,respectively construct a text classifier,and use accuracy,recall,F1 value and ROC The curve evaluates the performance of each classifier,and through experimental comparison,the optimal text classifier is used to classify the data for sentiment.(4)Application examples of "African Swine Fever".Food safety information is the starting point of this study.The data comes from the public opinion system of China's Jilin.com,and "African swine fever" is used as an example to study the sentiment tendency.Since the public emotions in this kind of data are mainly negative emotions and less positive emotions,this article innovatively proposes an emotional value calculation strategy based on "negative emotion intensity" to construct a time series-based emotional tendency chart.In addition,the public 's discussion points or concerns about the “African swine fever” incident in each time period are different,and the hot word statistics based on the time period are drawn to dig out the entity information in the public mood very intuitively.Through the above research,it can play a certain role in guiding the negative emotions in society and cultivating positive emotional preferences.