Implementation of Naive Bayes Algorithm for Spam Comments Classification on Instagram
Beta Priyoko, Ainul Yaqin · 2019 International Conference on Information and Communications Technology (ICOIACT) · 2019
Instagram is one of the popular social media in Indonesia, even in the world. With instagram, users can share their moments of life in the form of photos or videos. Instagram users can follow each other. But when a user already has a lot of followers, many instagram accounts also respond to posts with comments that can be categorized as spam. Spam comments are usually found in every account post that has a lot of followers, especially public figures in Indonesia and of course this is very annoying. Instagram has provided services to delete or hide comments, but a model is still needed to detect comments that are spam or notspam.The Naive Bayes algorithm will look for the probability of each class when the comments are inputed. Before the comments probability are calculated for each class, comments will be processed through the preprocessing stage, namely casefolding, cleaning, tokenizing, and stemming. After knowing the probability value of each class, then the probability value will be compared. If the highest probability value is a comment that is hypothesized as spam class, then the comment will be labeled as spam. If the highest probability value is a comment that is hypothesized as notspam class, then the comment will be labeled as notspam.Model test result : F1-measure of 0.83,recall of 0.98 and, precision of 0.72, It can be concluded that the classification of spam comments on Instagram in this research was successful.