Analyzing Students' Emotional Tendencies in the Context of Civic Education by Combining the Parsimonious Bayesian Approach

Abstract

The study applies a plain Bayesian algorithm classification model with LDA topic model and feature-weighted fusion algorithm for the analysis of students' sentiment tendency in the context of different Civic and Political Education topics. Firstly, the text of Civic and political education is mined through word splitting, feature selection, weight replication, and combining the TF-IDF algorithm with the LDA model; secondly, a feature-weighted fusion of the plain Bayesian sentiment classification algorithm is proposed to realize the optimization of the sentiment classification, comparing the keyword TF and IDF values to refine the textual information, and to deepen the students' understanding of the educational content; the use of the LDA model is applied in the theme of the Civic and political education content mining, and then analyze the student comment data for sentiment tendency. The results show that the algorithm proposed in this paper is better than other sentiment classification algorithms, with an accuracy of 95.92% on two-category data and 82.76% on multi-category data. Eight topic types were extracted using LDA model. Finally, the students' affective tendency was calculated, and the affective quantity and affective value of the affective features and themes in each stage of different contexts were output, which verified the accuracy and validity of the model's classification.