Corpus-Based Data Acquisition and Topic Analysis of Chinese-Related Public Opinion in Western Media

Abstract

This paper sorts out and discusses the different methods used in the analysis of China-related public opinion, builds a theoretical system of public opinion analysis methods on this basis, conducts case analysis and research on hot public opinion combined with corpus, and selects the semantic analysis method from the technical level supporting the public opinion analysis method. From the perspective of related technologies involving intelligent analysis methods, comparative analysis and improved applications are given, which provides an effective analysis basis for government public management departments to fully grasp and guide China-related public opinion. Sentiment tendency analysis is also an important direction of Chinese-related public opinion research. This paper sorts and analyzes China-related public opinion from the characteristics, applications, and text processing methods of different granularities and builds and improves the annotation model for the smallest granularity. The frequency, trend, and evolution characteristics of China-related public opinion events are sorted out and analyzed, and the trend analysis method is used to analyze the evolution trend of attention of public opinion events. The distribution of individual opinion acceptance, trust threshold, and opinion leaders are simulated by experiments. The impact on the evolution of China-related public opinion was determined. The experimental results show that the effect of the method proposed in this paper is improved by about 10% compared with the direct use of statistical learning methods for emotional orientation analysis. The SVM method also has obvious advantages over the Bayesian method, especially the combination of the SVM method and the bigram method which gives the best results.