SENTIMENT ANALYSIS USING K-NEAREST NEIGHBOR BASED ON PARTICLE SWARM OPTIMIZATION ACCORDING TO SUNSCREEN’S REVIEWS

Abstract

High UV exposure and tropical climate are afftected by the equator that passing through Indonesia. In this case, early aging will haunted not only by skincare lovers but also the rest of Indonesians including male and female. By so, we need to protect our skin using sun protector like sunscreen. Sunscreen application awareness through reviews by many different user probably are the most effective way to get to know the suitable sunscreen. By scrolling through the reviews surely will be time wasting. As of, sentiment analysis is the solution to classifying between negative and positive sentiments from the reviews. This research uses K-Nearest Neighbor (K-NN) algorithm as classification method because this method way more easy and efficient to use by its self learning, thus K-NN can learned its own data through its neighbor. Particle Swarm Optimization is used to increasing the accuration. Evaluation method using Confusion Matrix and the results are accuracy, precision and recall. Classification result using only k-NN and optimized with PSO are getting increase. Brand ‘Azarine’ rank the first accuration value with 91,89% using k-NN and 92,80% using k-NN PSO.