Analisis Sentimen Review Produk Skincare Dengan Naïve Bayes Classifier Berbasis Particle Swarm Optimization (PSO)
Tri Astuti, Yuli Astuti
Abstract
Skin care products have become the main needs of all people who are the targets of various brands of skin care products. However, not all skin care products have good quality according to consumer needs. They look for products that have the best quality by looking at reviews from other people, so they have an idea that influences their interest from other people's reviews submitted through various marketplace platforms or social media regarding the results after using these skin care products. Sentiment analysis is one way to analyze and classify reviews into positive opinions and negative opinions regarding the product in question to look for product quality based on public views. The algorithm used in this research is the Naive Bayes Classifier. The Naive Bayes Classifier method was chosen for reasons of ease of implementation, fast and high accuracy. The Naïve Bayes method also has a disadvantage, namely it is sensitive to feature selection, which results in low classification accuracy. Therefore, in this study, the feature selection method, namely Particle Swarm Optimization, was used in order to increase the accuracy of the Naïve Bayes classifier. The dataset used is 800 data reviews and tested using 10-Fold Cross Validation. The results showed an increase in accuracy from 77.96% to 79.85%.