Ideological opinion clustering identification based on Gibbs sampling in social new media environment

The effect of ideological public opinion in social opinion is growing as social new media develops rapidly. Effective mining of crucial information from the vast social new media data has become a hot issue in present opinion analysis study. Thus, this work presents an ideological opinion analysis model, GibbsCluster, derived from the combination of Gibbs sampling and K-means clustering. Using Gibbs sampling, the model separates opinion data into groups by means of the K-means and combines with sentiment analysis for fine-grained opinion classification. In the combined effect of opinion clustering and sentiment analysis, the experimental findings reveal that the GibbsCluster model much beats conventional approaches. This work also tests the adaptability of the model in other social platforms and creates a creative evaluation approach to fully evaluate its performance by accuracy and F1-score.