Komparasi Kinerja DIETClassifier dalam Klasifikasi Intent Data Sejarah Dewi Durga dengan Menggunakan Framework Rasa

Anggraeni, Dea Puspita (2025) Komparasi Kinerja DIETClassifier dalam Klasifikasi Intent Data Sejarah Dewi Durga dengan Menggunakan Framework Rasa. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 21082010029-cover.pdf Download (1MB)
	Text (Bab 1) 21082010029-bab1.pdf Download (268kB)
	Text (Bab 2) 21082010029-bab2.pdf Restricted to Repository staff only until 24 July 2027. Download (440kB)
	Text (Bab 3) 21082010029-bab3.pdf Restricted to Repository staff only until 24 July 2027. Download (1MB)
	Text (Bab 4) 21082010029-bab4.pdf Restricted to Repository staff only until 24 July 2027. Download (3MB)
	Text (Bab 5) 21082010029-bab5.pdf Download (256kB)
	Text (Daftar Pustaka) 21082010029-daftarpustaka.pdf Download (187kB)

Abstract

Historical data has an irregular structure and contains deep symbolic meaning, requiring an appropriate approach to process and classify it. This research applies the DIETClassifier model to historical data of Dewi Durga using the Rasa framework to classify intents in narrative texts. The approach used is CRISP-DM, starting with literature study, followed by data understanding and preparation. The data processed with NLP is then labeled with intents and divided into training and testing data. The DIETClassifier model is trained with nine different hyperparameter configurations. The best configuration, which uses 100 epochs, constrain similarity set to true, fine-tune set to true, batch size 16, learning rate 0.001, adamax optimizer, and dropout rate 0.2, achieves 99.63% accuracy and 99.57% F1-score on test data, demonstrating the model's best performance in intent classification. Evaluation is conducted using confusion matrix, accuracy, precision, recall, and F1-score to measure model performance. After training, the model is deployed locally and integrated into a web-based application. This research contributes to the development of NLP-based chatbots for preserving and presenting historical and cultural information more efficiently, while opening opportunities for further development in cultural text analysis using NLP technology.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Wahyuni, Eka Dyar	NIDN0001128406	ekawahyuni.si@upnjatim.ac.id
Thesis advisor	Suryanto, Tri Lathif Mardi	NIDN0025028902	trilathif.si@upnjatim.ac.id

Subjects:

Q Science > QA Mathematics > QA76.6 Computer Programming
T Technology > T Technology (General) > T58.6-58.62 Management Information Systems

Divisions:

Faculty of Computer Science > Departemen of Information Systems

Depositing User:

Unnamed user with email 21082010029@student.upnjatim.ac.id

Date Deposited:

25 Jul 2025 01:31

Last Modified:

25 Jul 2025 01:31

URI:

https://repository.upnjatim.ac.id/id/eprint/40771