Tendency Analysis of Thematic Networks Public Opinion Based on Sentiment Ontology

The algorithms of tendency analysis based on text classification, semantic rules pattern and sentiment word are designed based on the study of the text tendency recognition method. The combination of the sentiment ontology construction and the emotional concept selection method based on HowNet and subject areas corpus can improve the concept comprehensiveness and domain pertinency of sentiment ontology. The keywords are extracted and their degrees of emotional tendencies are determined by sentiment Ontology. The degree of emotional tendencies is used as the characteristic weight combining the syntactic rules and the adverbs of the sentiment. The emotional analysis of web texts for thematic networks public opinion is done by machine learning methods. The experiments show that the analysis results has higher precision and recall. The universality and stability of its implementation deserves is worth of further study.