KOMBINASI TEKNIK AUGMENTASI DAN MODIFIKASI ARSITEKTUR TRANSFORMER DALAM MENDETEKSI DEPRESI PADA DATASET TWITTER BAHASA INDONESIA

SALSABILA, AULIA and Suprihatin, Bambang and Desiani, Anita (2023) KOMBINASI TEKNIK AUGMENTASI DAN MODIFIKASI ARSITEKTUR TRANSFORMER DALAM MENDETEKSI DEPRESI PADA DATASET TWITTER BAHASA INDONESIA. Undergraduate thesis, Sriwijaya University.

	Text RAMA_44201_08011281924049.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (2MB) | Request a copy
	Text RAMA_44201_08011281924049_TURNITIN.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (8MB) | Request a copy
	Text RAMA_44201_08011281924049_0026017102_0011127702_01_front_ref.pdf - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (1MB)
	Text RAMA_44201_08011281924049_0026017102_0011127702_02.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (453kB) | Request a copy
	Text RAMA_44201_08011281924049_0026017102_0011127702_03.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (277kB) | Request a copy
	Text RAMA_44201_08011281924049_0026017102_0011127702_04.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (769kB) | Request a copy
	Text RAMA_44201_08011281924049_0026017102_0011127702_05.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (58kB) | Request a copy
	Text RAMA_44201_08011281924049_0026017102_0011127702_06_ref.pdf - Bibliography Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (184kB) | Request a copy
	Text RAMA_44201_08011281924049_0026017102_0011127702_07_lamp.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (153kB) | Request a copy

Abstract

Depression is a mental health disorder that can affect a person's activities in leading a useful life. One way to detect depression on social media, such as Twitter, is by classifying text. In the text classification process, a large amount of text data is needed, but the text data available in Indonesian is still limited, so augmentation techniques are applied to increase the amount of text data. Easy Data Augmentation (EDA) and back translation are some of the commonly used augmentation techniques. Augmentation data can be processed using a certain architecture. One of the architectures that can be used in processing text data is Transformer. The advantages of Transformer, which is able to process all the words at one time simultaneously. In this study, augmentation was carried out on the Indonesian language Twitter dataset using a combination of EDA augmentation techniques and back translation. After augmentation, classification is carried out using the Transformer architecture. Several stages were carried out in this study, namely data augmentation, text preprocessing, tokenization, pad sequences, training, testing, evaluation of results, and conclusions. The results obtained in this study are accuracy, precision, recall, and f1-score above 97% with a very good category. This shows that the Transformer architecture has a good performance in classifying text in detecting depression.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Back Translation, Depresi, Easy Data Augmentation, Klasifikasi, Transformer
Subjects:	Q Science > QA Mathematics > QA8.9-QA10.3 Computer science. Artificial intelligence. Computational complexity. Data structures (Computer scienc. Mathematical Logic and Formal Languages
Divisions:	08-Faculty of Mathematics and Natural Science > 44201-Mathematics (S1)
Depositing User:	Aulia Salsabila
Date Deposited:	28 Feb 2023 02:54
Last Modified:	28 Feb 2023 02:54
URI:	http://repository.unsri.ac.id/id/eprint/90091

Actions (login required)

View Item