<![CDATA[User reviews of the app, including popular games like Clash of Clans, are an important indicator of the quality of the app. However, fake reviews can mislead users and damage the app's reputation. This study aims to classify Clash of Clans reviews on the Google Play Store as "valid" or "invalid" using three algorithms: decision tree (DT), K-Nearest Neighbors (KNN), and Multinomial Naive Bayes (MNB). The dataset used contains 320 Indonesian reviews with a rating of 1 to 3 and validated by experienced players. Text features are extracted using TF-IDF and the model is evaluated using cross validation. The results show that the decision tree has the highest accuracy (64%), followed by MNB (59%) and KNN (53%). Cross validation shows that MNB has the most stable performance, while KNN is more sensitive to data changes. Decision Trees show the lowest performance and are less effective on new data because they tend to overfit. The study provides valuable insights into the selection of user review classification algorithms by considering accuracy, precision, acquisition, and performance stability.]]>
Copyrights © 2024
