Comparison of Filter and Wrapper Based Feature Selection Methods on Spam Comment Classification

The continuous growth of the internet has led to use social media for various purposes increase. For instance, some irresponsible parties take advantage comment feature on platforms harm others by providing spam comments shared object. Furthermore, variation creates many features be processed, thereby negatively impacting performance a classification algorithm. Therefore, this study aims solve problem associated with comparing filter and wrapper based selection using text techniques. Data collected from training test data 4944 100 showed that best accuracy, precision, recall, f-measure MNB are 96%, 100%, 92%, 95.8%. accuracy is achieved combining Chi-Square Sequential Forward Selection methods subset 500 features. increase in SVM classifications 8% 4%. This research concludes combination improves Indonesian language comments.