<![CDATA[Social media is a socializing activity through the internet, which has many conveniences that allow people to communicate and access information quickly. Social media is widely used to get news that is difficult to get. Social media applications are quite popular such as Twitter and recently social media that has similar features, namely Threads. Therefore, the purpose of this study is to compare 3 algorithmic methods of user-generated reviews on two applications, namely Twitter and Threads. We used 899 reviews on Twitter, with 245 positive sentiments and 654 negative sentiments, and 638 reviews on Threads, with 220 positive sentiments and 418 negative sentiments. Cleansing, preprocessing, and modeling are the steps that will be passed to process the data. In this study, split data and cross validation models were used, and the three algorithms used were Naïve Bayes, Decision Tree, and KNN, with a ratio of 80:20 for training data and test data. The accuracy value obtained for Naïve Bayes is 85.56%, Decision Tree is 72.78%, and KNN on the Twitter application, while the threads application gets 66.41% for Naïve Bayes, Decision Tree gets 65.41%, and the threads application gets 66.41%. In the Naïve Bayes algorithm, precision, recall are calculated in the Threads and Twitter applications. The Threads application gets 64.86% in precision and 73.85% in recall, while the Twitter application gets 84.69% in precision and 88.30% in recall.]]>
Copyrights © 2023
