FAKE REVIEW DETECTION ON E-COMMERCE SITES USING MACHINE LEARNING

• Main Author: RIZDAPUTRA (NIM : 13513027), AHMAD
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/25200
• Institution: Institut Teknologi Bandung

Existing reviews on e-commerce sites are taken into consideration by prospective buyers to purchase a product. Thus, some companies try to manipulate some reviews. These reviews are called fake reviews. Existing research has been focused on features that are on one point of view, only textual features or other features plus a few textual features. In this paper, we combine the features from previous research. There are 37 features implemented and 4 algorithm i.e. logistic regression, support vector machine (SVM), decision tree, and voting. The experiment uses around 500 thousands of data from amazon.com. There are 18 experiments conducted in this paper, 10 are based on the amount of training data and 8 are based on the types of review. The results are evaluated with F1-score, area under the ROC curve (AUC), and confusion matrix. The result are not bad and decision tree algorithm give the best result and outdoes other algorithm.