PERBANDINGAN SUPPORT VECTOR MACHINE DAN K-NEAREST NEIGHBOR PADA ANALISIS SENTIMEN ISU COVID-19 TERHADAP KEMENKES RI

In the current pandemic situation, social media is the right place to spread information like Instagram. Like the Indonesian Ministry of Health, the ministry has an Instagram account that is used to share information such as notifications and appeals related to Covid-19. In this post, there are a lot of comments written by the public to express their opinions about the policies that have been set in order to deal with the Covid-19 pandemic. However, public opinion on the post is very diverse, the diversity of opinions listed, if not managed, will cause provocation, causing distrust of the Covid-19 virus to the point of loss of trust in the government in carrying out its duties. Therefore, this study proposes sentiment analysis to find out public opinion with the SVM method which is classified into three classes, namely positive, negative and neutral. Then, it is compared with the k-Nearest Neighbor (k-NN) method which aims to find out the best method between the two. The data that was successfully retrieved were 2034 valid data, then went through stages such as case folding, tokenizing, stopword removal, normalization, stemming, and TFIDF weighting. The result is that the SVM method gives higher accuracy than k-NN, which is 98.30% at C=0.8 and the RBF kernel.