Sentiment Analysis of the Use of Makeup Products Using the Support Vector Machine Method

Abstract

Many beauty products have emerged from various brands by providing attractive offers for women who are their main targets. Product reviews can help consumers regarding the quality of using the product. However, the problem is, on the femaledaily.com website there is no distinction between negative, neutral, and positive reviews so that consumers must first read the review and it takes a lot of time and this problem really requires a classification process on the review into negative, neutral, and positive classes. This process cannot be done automatically, therefore sentiment analysis is needed. To find out the classification of positive, negative, and neutral sentiment on the product, the Support Vector Machine (SVM) method is used, the advantage of SVM in this case lies in its ability to handle high-dimensional datasets and still produce effective classification and SVM is also a good choice for sentiment analysis in the context of cosmetic product reviews. The classification results using the SVM method produce data into 3 classes, namely 510 positive reviews, 98 neutral, and 29 negative with an accuracy value of 77.97%, precision 78%, recall 100%, fi-score 88%.