Using Sentiment Analysis to Analyze the Feedback of Students with Open-ended Questions

Abstract

This study measures students’ preferences by analyzing their feedback about the lectures they have been attending based on open-ended questions on the course evaluation using the sentiment analysis method, one of the text mining methods. Based on this analysis, the quality of lectures was measured to suggest future changes made for improving teaching methods, directions for improving lecture quality, and its contents. This study aims to analyze qualitatively the feedback of student evaluation to examine student’s perceptions about the lecture more directly and comprehensively. To this end, answers of open-ended questions were collected and classified with positive, neutral, negative opinions or polarity. As a result of sentiment analysis 422 words are extracted. Among them 327 words (82.2%) were positive as opposed to 44. To measure our model’s accuracy, recall and precision and all of them are over 0.96 which indicates that machine learning model using NBC method successfully classified the data. This study could contribute to the course evaluation because it shows how to utilize students’ feedback of open-ended questions using text mining and sentiment analysis. It also gives solid measurement of the subjective point of view of students.