<![CDATA[The presence of Twitter as an online forum causes everyone to be free to comment. This is one of the reasons the government issued the law on Electronic Information and transactions (ITE) to oversee all activities in cyberspace. However, the growing and growing amount of comment data is also increasingly difficult to analyze. Therefore, the application of data mining using the Rapid Miner application is proposed in this study to help in finding the most effective and efficient method, by looking at the accuracy, precision, and time lapse required when processing the comment data. In this study, a total of more than 12,000 data, containing comments and seven types of labels based on ITE were collected for analysis. After pre-processing the data, the researchers chose five of several classification methods, namely Naive Bayes, k-NN, Decision Tree, SVM, and Perceptron methods to be tested. From the tests that have been carried out through the Rapid Miner application, it was found that SVM became the best method used to classify comment data, with an accuracy of 55.32%. Meanwhile, the method with the lowest accuracy is occupied by Perceptron with a total accuracy of only 18.04%. Based on observations, the best accuracy results only reached 55.32% due to the large number of labels considered in the prediction.]]>
Copyrights © 2024
