Improving course review helpfulness prediction through sentiment analysis
Fetty Fitriyanti Lubis, Yusep Rosmansyah, Suhono Harso Supangkat · 2017
Participant reviews from course website have valuable information for candidate participants and course providers The reviews help other participants interpret whether the course will be enrolled or not. However, reviews have variations in both quality and usability. Course aggregator website has voting systems where participants can vote on whether or not a review is helpful to them. For a popular course, the number of reviews can be thousands. Thus, users have difficulty filtering out useful information. In this paper, program prototypes are developed to categorize helpful reviews automatically. It was able to automatically classify helpful reviews using the Naive Bayes algorithm. Our experiments show that performing data filtering with sentiment analysis can improve the performance of evaluation parameters in classifying useful reviews.