USER SENTIMENT ANALYSIS METHODS FOR ELDERLY SOCIAL MEDIA NETWORKS

For the analysis of user sentiment in social media networks for the elderly population, emotional sentences are first extracted to classify movie reviews. Afterwards, social network data of the elderly population based on user search behavior is analyzed. The movie reviews of elderly social media users are analyzed for rating prediction. The research results indicate that the accuracy of sentiment classification results is in descending order of Dirichlet, maximum entropy, and support vector machine. The highest classification accuracy of the three algorithms is 87.1%, 86.9%, and 86.5%, respectively. The classification accuracy of the first level classifiers of Dirichlet, maximum entropy, and support vector machine are 90.7%, 88.7%, and 87.4%, respectively. The classification accuracy of the second level classifier is 86.7%, 83.7%, and 80.4%, respectively. The predictive analysis results of the research method are superior to those generated by using Slope One. The method proposed in the study can promote emotional analysis of film review texts, improving the analysis accuracy.