Text Sentiment Analysis Of Public Opinion Events Based On Multi-source Data

With the rapid development of the mobile Internet,social networks have become an important platform for users to obtain information,express opinions,and communicate.Network users share their findings through texts,pictures,short videos and other forms to express their opinions and positions..In recent years,with the diversification of social media platforms and the enrichment of network content,whenever major public opinion events occur,they will be fermented and fermented on various social platforms,which has become a hot topic in the entire network.Especially in some negative public opinion events,if they are not guided and controlled in a timely manner,they will easily lead to mass or violent events,and even endanger the health and stability of the lives and property of the people and the healthy and stable development of social harmony.Based on this,from the perspective of multi-language dimension and multiple data sources,this paper conducts sentiment analysis of the network public opinion text from the cyberspace environment of both parties of the event,and strives to more comprehensively analyze the network users' emotional attitude towards the public opinion event.First,the thesis introduces and analyzes the relevant concepts of network public opinion,text sentiment,and multi-source data,and evaluates the current research status at home and abroad in conjunction with the current Internet development form.Based on the results of literature research,this paper proposes the framework theory and research ideas of this paper.Secondly,construct a sentiment analysis model of network public opinion text based on multi-source data according to the data source of the collection layer,the data source of the processing layer and the data source of the decision layer.Model subdivision data collection,text preprocessing,sentiment dictionary generation,sentiment intensity calculation,sentiment topic clustering and other modules provide theoretical basis and practical guidance for the following practice.Finally,taking the "China-US Trade War" as a research case,it is processed according to the process of the multi-source data sentiment analysis model,and the text corpus on Twitter and Weibo is collected to obtain the overall sentiment trend and the main sentiment factors.According to the only Chinese corpus,Only the English corpus and the mixture of Chinese and English are used for comparative analysis to reveal the difference between the results of multi-source data and single data analysis.The results show that:(1)Model analysis level: The effect of online public opinion analysis based on multi-source data is more comprehensive and objective than that of a single data source;(2)Core level of the event: The core issue of the Sino-US trade war is trade,Game between market and economy;(3)Granularity of text sentiment: Domestic users have a positive sentiment towards the Sino-US trade war,but most of the Twitter texts show a negative sentiment..