DSpace logo

Please use this identifier to cite or link to this item: https://repository.uksw.edu/handle/123456789/19553
Title: Analisis Sentimen terhadap Akun Penyebar Ujaran Kebencian di Twitter
Authors: Zakharia, Restu
Keywords: Twitter;analisis sentimen;multilayer perceptron;DeepDetect®
Issue Date: 2019
Publisher: Program Studi Sistem Komputer FTEK-UKSW
Abstract: Social media has become popular for Indonesian people, including children, teenagers and adults. Social media, Twitter for example, is used to disseminate information to public, which then can be easily replied by others. Of course, there might be positive or negative responses. Based on that reason, a tweet detection system containing hate speech on Twitter using DeepDetect® Server is proposed. The design of this system aims to classify tweets containing hate speech on Twitter based on tweets that have been uploaded. This classification is divided into 2 types of classes, namely hate speech and neutral. The testing phase is done by entering input in the form of text/tweet sentences, tweet URLs and Twitter account names. The final result of the testing of the system in this paper is the percentage of tweets containing hate speech and neutral. After conducting several training processes using the multilayer perceptron method, the best machine learning model was obtained with 50 hidden neurons in 1 layer using relu activations, learning rate of 0.009, batch size of 64, test split of 10% and iterations of 1000. From the training results obtained the best model with precision of 90.42 %, recall of 90.30 % and accuracy of 90.15 %. Tests were carried out on 31 tweets and 6 different Twitter accounts. From 31 tweets, 93.55% successed to detect correctly and 6.45% were incorrect. The incorrect prediction results are due to the lack of the number of datasets used in this thesis. The percentage of success can be increased by adding more valid datasets so that predicted tweets can vary.
URI: https://repository.uksw.edu/handle/123456789/19553
Appears in Collections:T1 - Computer Systems

Files in This Item:
File Description SizeFormat 
T1_622014004_BAB I.pdfBAB I150.69 kBAdobe PDFView/Open
T1_622014004_BAB II.pdfBAB II974.21 kBAdobe PDFView/Open
T1_622014004_BAB III.pdfBAB III623.24 kBAdobe PDFView/Open
T1_622014004_BAB IV.pdfBAB IV1.29 MBAdobe PDFView/Open
T1_622014004_BAB V.pdfBAB V97.53 kBAdobe PDFView/Open
T1_622014004_Daftar Pustaka.pdfDaftar Pustaka69.67 kBAdobe PDFView/Open
T1_622014004_Judul.pdfHalaman Judul2.23 MBAdobe PDFView/Open
T1_622014004_Lampiran.pdfLampiran487.14 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.