Sentiment analysis on feedback of higher education teaching conduct: An empirical evaluation of methods
Jimmy Jimmy, Vincentius Riandaru Prasetyo · AIP conference proceedings · 2022
Sentiment analysis aims to automatically identify and classify the tone or polarity of people’s opinion in an unstructured text. This paper focuses on evaluating the accuracy of sentiment analysis methods to classify Indonesian higher education teaching conduct. In this context, we make the following contributions: (1) evaluation on the impact of text preparation methods in term of accuracy of sentiment analysis results, (2) evaluation the accuracy of three popular sentiment analysis methods (i.e., Naive Bayes, Support Vector Machine, Decision Tree) in classifying Indonesian text, and (3) proposal and evaluation on the effectiveness of combining the results from the three methods considered in this study with hope to improve the results’ accuracy. Finally we analyzed cases where all classifiers suggested incorrect sentiment classification and highlighted areas for future works to improve the accuracy of sentiment analysis, in particular for Indonesian Text.